BEGIN:VCALENDAR VERSION:2.0 PRODID:-//HasGeek//NONSGML Funnel//EN DESCRIPTION:India's most renowned data science conference X-WR-CALDESC:India's most renowned data science conference NAME:The Fifth Elephant 2016 X-WR-CALNAME:The Fifth Elephant 2016 REFRESH-INTERVAL;VALUE=DURATION:PT12H SUMMARY:The Fifth Elephant 2016 TIMEZONE-ID:Asia/Kolkata X-PUBLISHED-TTL:PT12H X-WR-TIMEZONE:Asia/Kolkata BEGIN:VEVENT SUMMARY:Check-in and breakfast DTSTART:20160728T030000Z DTEND:20160728T035000Z DTSTAMP:20260421T155209Z UID:session/9Y2uh9AB2wWjbiTPYbx7Ru@hasgeek.com SEQUENCE:0 CREATED:20160606T050602Z DESCRIPTION:\n LAST-MODIFIED:20160606T050608Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Check-in and breakfast in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Introduction to The Fifth Elephant 2016 DTSTART:20160728T035000Z DTEND:20160728T040000Z DTSTAMP:20260421T155209Z UID:session/578vFomMbw6KNAELKHY8CQ@hasgeek.com SEQUENCE:0 CREATED:20160606T050627Z DESCRIPTION:\n LAST-MODIFIED:20200619T062516Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Introduction to The Fifth Elephant 2016 in Auditorium 1 in 5 m inutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Using Data to Identify the Genomic Cause of Disease DTSTART:20160728T040000Z DTEND:20160728T050000Z DTSTAMP:20260421T155209Z UID:session/GVLr24pj61PD7ZdxHMx6qN@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160722T075956Z DESCRIPTION:.\n\n### Speaker bio\n\nRamesh Hariharan is CTO at Strand Life Sciences and Adjunct Professor at the Indian Institute of Science. For mo re information\, see http://www.hariharan-ramesh.com\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/using-data-to-identify -the-genomic-cause-of-disease-GVLr24pj61PD7ZdxHMx6qN BEGIN:VALARM ACTION:display DESCRIPTION:Using Data to Identify the Genomic Cause of Disease in Auditor ium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Allocation and Forecasting in Guaranteed Delivery of Advertisement s DTSTART:20160728T040500Z DTEND:20160728T045000Z DTSTAMP:20260421T155209Z UID:session/6FzJn95s6E9tSEvYchmXLW@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160719T092415Z DESCRIPTION:In this talk\, we will discuss the following: \n\n1. Allocatio n: In GD\, as advertisements don't compete on bids\, we allocate advertise ments to user-views such that we deliver on the guarantees\, while keeping the advertiser's interests (reaching the right set of users) intact. This problem is modelled as constrained optimization over bipartite graph of a dvertisements and audience segments. \n2. Forecasting: We need to know the number of views in the future that we get from different audience-segment s. We model views in an audience segment\, as a time series and various ex ternal inputs as exogenous variables which affect the time series. We will briefly describe the various algorithms and processes that we follow to e nable forecasting.\n\n### Speaker bio\n\nAditya Rachakonda is a Data Scien tist at Flipkart Ads. He is currently working on problems in advertisement optimization. Earlier\, he was a Research Scientist at Big Data Labs in A merican Express. He has experience in machine learning\, text mining and g etting insights from short and noisy texts. Aditya is a PhD in Computer Sc ience from IIIT Bangalore and his research interests include semantics in text and information retrieval.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/allocation-and-forecas ting-in-guaranteed-delivery-of-advertisements-6FzJn95s6E9tSEvYchmXLW BEGIN:VALARM ACTION:display DESCRIPTION:Allocation and Forecasting in Guaranteed Delivery of Advertise ments in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Keynote: The Alternative Data revolution on Wall St DTSTART:20160728T050000Z DTEND:20160728T060000Z DTSTAMP:20260421T155209Z UID:session/Sog8Yhe61x7CANbpRsTxzB@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160715T115212Z DESCRIPTION:Topics covered will include aspects of the developing alternat ive data ecosystem including:\n\n* Alternative data R&D process flow\n* Co mputing infrastructure and the technology stack\n* Research & analytics pr oviders\n* Technical solutions to common issues found in alt. data\n* Best practices\n\nWe're going to walk through a few examples of how noisy\, un structured data become an investable signal using tools such as text minin g and machine learning. The aim is to introduce the audience to the proces s of how hedge fund portfolio managers and sell-side research analysts are systematically generating returns by leveraging unique primary (bots / sc rapers\, channel checks) and third party datasets (including data brokers) . This includes sourcing\, compliance\, scrubbing out PII\, alpha generati on related to revenue estimates and approaches to balance the secret sauce with product transparency. Finally\, we'll ponder the future of alternat ive data in finance and touch on how companies in the data space can best take advantage of this growing trend.\n\n### Speaker bio\n\nGene Ekster wa s previously head of R&D at Point72 Asset Management (formerly SAC Capital )\, a Director of Data Product at 1010Data and a Senior Analyst at Majesti c Research (now ITG Investment Research). Currently\, Gene works with asse t management firms and data providers in a consulting capacity to help int egrate alternative data into the investment process. He can be reached via LinkedIn (https://www.linkedin.com/in/geneekster).\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/keynote-the-alternativ e-data-revolution-on-wall-st-Sog8Yhe61x7CANbpRsTxzB BEGIN:VALARM ACTION:display DESCRIPTION:Keynote: The Alternative Data revolution on Wall St in Auditor ium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Morning tea break DTSTART:20160728T060000Z DTEND:20160728T063000Z DTSTAMP:20260421T155209Z UID:session/NPiRkqt6svorW6F2tumo2Y@hasgeek.com SEQUENCE:0 CREATED:20160606T050900Z DESCRIPTION:\n LAST-MODIFIED:20160720T090238Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Morning tea break in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:ML in fin-tech - Transforming 60 crore Indian lives DTSTART:20160728T063000Z DTEND:20160728T071500Z DTSTAMP:20260421T155209Z UID:session/DtzWBzjNVAXaby3qM3V4Z2@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Beginner CREATED:20160715T114839Z DESCRIPTION:- Proof that Banking is the most data-intensive business in th e world!\n- How will big-data help increase penetration of credit in India \n - How to increase penetration of credit in India\n - The Long tai l\n - Using big-data to personalise\n - The future of Holistic Risk- assessment and Differentiated risk-pricing\n- A sneak-peak into our Credit -scoring-engine architecture\n- Examples of how to use AI and ML in every imaginable way to provide *access to credit* to people in the long tail in India - over 60 crore people who would otherwise be rejected by any Bank or Financial Institution\n - E.g. image recognition\n - information retrieval and NLP\n - deep learning\n - social network analysis\n - fraud detection and prevention\n - The importance and impact of Re-i nforcement learning\n- Examples of why ML is harder on these ever-evolving dynamic datasets.\n\n### Speaker bio\n\n- I studied BS and MS at **Stanfo rd University** in Computer Science\, built GraphSearch with the creator o f Google Maps Lars while working as a Facebook engineer\, was a VC associa te at MDV ($700M fund) post that\, and the youngest PM at Microsoft HoloLe ns post that\, among other experiences. My two specialisations at Stanford were Systems and AI. I also enjoy behavioral psychology and we're employi ng it in our product\, design\, and risk-assessment algorithms.\n- I am co -leading the largest consumer-lending fin-tech company in India:\n - Fi n-tech in India is a trillion$ market because of India’s size. Today we have 25 crore people online and in the next 5 years that number is poised to grow to around 70-80 crore. This kind of a relative and absolute growth will not be witnessed again in the world\n - India has ~53% of its pop ulation under 25 years of age. It is a country of millennials\, for whom\, banking and lending have to be re-imagined.\n - India is mobile-first. \n- Finomena\, the company I co-lead\, is bringing Financial Inclusion to a country of 1.25 Billion people through focusing on the holy-grail of “ small tenure\, small ticket size” lending which can only be done in a te ch-first way to be effective.\n- Finomena is built on the foundation provi ded by Aadhaar (the world's largest biometric fingerprint and identity pla tform of it's kind with 1 Billion people already registered on it). The Aa dhaar platform will be as revolutionary over the next 5 years as the smart phone platform was from 2010-2015 in India.\n- I am a contributor to the “India-Stack” discussions - this refers to the most advanced fin-tech infrastructure in the world being developed by the Indian govt and several leading private sector thought-leaders in India\n- India is it’s own be ast (because of it’s scale\, diversity and chaos)\, and requires it’s own innovations\, and I am in the rare position to compare and contrast th e happenings in India with those in US (because I spent 7 years there) and /or China today.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/ml-in-fin-tech-transfo rming-60-crore-indian-lives-DtzWBzjNVAXaby3qM3V4Z2 BEGIN:VALARM ACTION:display DESCRIPTION:ML in fin-tech - Transforming 60 crore Indian lives in Auditor ium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Purpose\, Speed & Visibility : Facilitating product discovery & en gagement on a e-commerce website DTSTART:20160728T063000Z DTEND:20160728T071500Z DTSTAMP:20260421T155209Z UID:session/CXt7gHH8hoVVHRhHbPprfr@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160719T100207Z DESCRIPTION:I will start with defining a user’s intent\, web journey and how products sell on an ecommerce website - defining the “discoverabil ity” and “engagement” goals. I will then introduce relevance\, recal l & performance/engagement based discoverability from a search ranking per spective.\nThough the two goals above(discoverability & engagement) look i nclusive\, I will showcase using the 3 problems\, how conventional metrics such as Bounce rate\, product view rate\, ATC rate\, Conversion rate etc. can mask serious challenges that end-users face - thus opening up opportu nities about a Merchant can actually address for its users. \n\n\nProblem # 1 : Understanding lost carts \nData from a merchant with abandoned carts - where we figured out it was a problem with stock-replenishment of popul ar sizes and people were using carts to “bookmark”. While we do not co ntrol/influence a merchandiser‘s inventory - how could we change “the search storefront” (aka the search results page) - to reflect the availa bility factor that blended the popularity factor - also introducing the su pply & demand dynamics at this point. \nIn the process\, we ended up buil ding a clustering solution for mapping sizes across different categories\, which then fed into the availability factor. \n\nProblem # 2 : Handling Special events - Mother’s day\, Back to school\, Halloween & holiday sal es \, special sports events etc. \, Marketplace products & New launches \n \nUsing data from the sudden redirect pages that a Merchant’s platform t eam sets up - to understand a Merchant’s Business goals\, quantitatively and then reflecting this in “discoverability” score I introduced befo re. This means that the evolving intent for these special events now ties to the hot products that merchandizer knows best will sell. \n Also\, all products are created unequal & by revisiting to a product's purpose - we c an proxy a user's intent. Since a product only exists in the realm of a “user intent” - bootstrapping some fair impressions is more challengin g than it looks. \n\nOne goal could be ensuring the new products have some impressions before they starve & set a downward spiral in the ecosystem\, but this when also coupled with non-sellable/zombie products that a merch ant introduces ahead of launches (eg iphone 6s\, Motorola next gen etc.) s kews the performance data. eg\, some products get a lot of impressions\, b ut do not sell since they are not sellable -this problem will showcase how do we flex the two uber goals of discoverability & engagement for product s.\n \nProblem # 3 : Understanding users segments on a website\nWe ca n't fix what we don't understand is broken. Aside of search\, users use v ery suggestive sort parameters\, dynamic filters\, paginatation - this pro blem will showcase data about "intent classification" that helped our sear ch tuning efforts. Brand sensitive vs. price sensitive users are a differe nt breed - and this has search storefront implications\, not just for what to show (close substitutes\,complementary goods\, related but neither sub stitutes\, not complementary goods)- but also on how we measure the intent 's differently. There is no such thing as an average user\, afterall.\n\n# ## Speaker bio\n\nEkta Grover is a Member Technical Staff & Data Scientist at Bloomreach Inc\, a firm that helps B2B and B2C companies make their co ntent more discoverable\, relevant & personalized\, while growing the bott om line for the Businesses. At Bloomreach she focuses on influencing propr ietary Search ranking signals in core search\, commoditizing analysis and shaping decisions that impact bottom line Business metrics for some of the largest retail merchants. \n\nShe has experience leading & building data tools in location based Mobile advertising\, e-commerce\, Search & Persona lization. Across her professional footprint\, she has worked with both Ent erprise product companies (SAP\, VMware) and midsize/small startups in con sumer web - and has helped shape products that affects million of users an d billions of dollars in incremental revenue for some of the largest e-com merce firms. She has a background in Quantitative Economics & Computer Sci ence.\n\nHer previous talks with HasGeek \nFifth Elephant Conference\, 201 4 : Experimentation to Productization : developing a Dynamic Bidding syst em for a location aware Mobile landscape\nPycon 2013 : Experiments in dat a mining\, entity disambiguation and how to think data-structures for desi gning beautiful algorithms\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/purpose-speed-visibili ty-facilitating-product-discovery-engagement-on-a-e-commerce-website-CXt7g HH8hoVVHRhHbPprfr BEGIN:VALARM ACTION:display DESCRIPTION:Purpose\, Speed & Visibility : Facilitating product discovery & engagement on a e-commerce website in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Taking Fashion and Lifestyle Commerce Towards SKUs Using Deep Imag e and Text Parsing DTSTART:20160728T071500Z DTEND:20160728T080000Z DTSTAMP:20260421T155209Z UID:session/M1b5SNzoCX61iL59WURmpC@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160719T100247Z DESCRIPTION:This talk will present recent innovations in deep neural netwo rks to build business applications using large scale data. We will take de ep dive into fashion and lifestyle online commerce data\, and image and te xt processing to build large-scale deep learning models.\n\nMotivation: Wh at is the quality of commerce catalogues and why product search experience is not satisfactory?\nHow often do you open multiple tabs to search for a n item? How many times you find a fashion item that you are looking for? D o the results match to your query intent? Most often not. To dig deeper\, we will first review some interesting statistics on the state of pollution in fashion and lifestyle commerce catalogues\; for instance\, number of d uplicate products\, number of products with missing keywords\, number of p roducts with mismatch between image and text. No wonder\, when you search for "blue evening cocktail party dress"\, you get poor results on most of the commerce platforms. We did this analysis on more than 10 million produ cts from different e-commerce portals in India and abroad.\n\nChallenges: Why has this state not been improved over the years?\nNormalizing and clea ning unstructured image and text data into a structured data poses several difficulties. Product images come in different size\, shape\, pose\, cont ent and other varieties. They contain different product items that may or may not be relevant to accompanying text. Text description contains mix of product description\, complementary products and suitability criteria. Pa rsing such images and text snippets on scale and with high accuracy has be en traditionally difficult for software machines (using machine learning a lgorithms). \n\nRebirth of deep learning to utilize big data in fashion co mmerce:\nI will first motivate why previous attempts of using machine lear ning to parse e-commerce data have not been entirely successful. I will th en describe what has changed with the rebirth of deep learning to solve th e problems of deep image and text parsing. Innovating and applying deep le arning models\, I will then show in details how we can extract structured data from unstructured image and text data to build SKUs for fashion and l ifestyle products. This category of products is especially challening to p repare SKUs since we have to extract a lot visual and textual attributes v ia deep parsing of images and text\; unlike consumer electronics category where product specifications are standardised.\n\nDeep Learning at scale:\ nI will then describe deep learning engineering pipeline to collect\, clea n and feed data to deep learning models\, train deep learning models using GPUs\, innovating on architectures and training procedures to achieve des ired accuracy\, and deploying models in production to clean and normalzie millions of products. I will especially talk about recent advances in full y convolutional and segmentation based deep neural networks and its applic ations for image and text processing at Infilect. \n\nTalk outline: https: //www.dropbox.com/s/1ph6fslz9wn3inu/fifth-elephant-2.pdf?dl=0\n\n### Speak er bio\n\nVijay Gabale is co-founder and CTO of Infilect\, an AI-powered C ommerce Platform. Infilect has been building a fashion commerce platform t o provide exceptional shopping experiences to the Internet consumers. The company has made several innovations in deep learning to process rich mult i-media data (text\, image\, videos) to improve discovery\, search and per sonalization experiences of online consumers. \n\nPrior to co-founding Inf ilect\, he was a research scientist with IBM Research Labs. He graduated w ith a PhD from IIT Bombay\, India in 2012. He has several top tier researc h publications and software patents to his name. He is also co-organizer o f 'Deep Learning Bangalore' meetup. He has been actively working in deep l earning for past several years and has give several talks in and outside I ndia on the research and applications of deep learning in e-commerce.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/taking-fashion-and-lif estyle-commerce-towards-skus-using-deep-image-and-text-parsing-M1b5SNzoCX6 1iL59WURmpC BEGIN:VALARM ACTION:display DESCRIPTION:Taking Fashion and Lifestyle Commerce Towards SKUs Using Deep Image and Text Parsing in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Deciphering Driving Behaviour using Geospatial Temporal Data Colle cted from Smartphone Sensors DTSTART:20160728T071500Z DTEND:20160728T080000Z DTSTAMP:20260421T155209Z UID:session/PJBiMVWBTz4uSzZCmmYPy7@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160718T123248Z DESCRIPTION:In this talk\, I will present 'What\, Why and How' of this tec hnology that we believe will bring smartphone-powered road safety to citie s\, fleets\, and individuals.\n\n### Speaker bio\n\nAditya Karnik is a dat a scientist at Zendrive Technologies\, Bangalore. Before coming to Zendriv e he had stints at General Electric and General Motors Research labs. He h olds PhD in ECE from Indian Institute of Science\, Bangalore. His research interests are in whatever it takes to solve important\, mostly applied\, problems.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/deciphering-driving-be haviour-using-geospatial-temporal-data-collected-from-smartphone-sensors-P JBiMVWBTz4uSzZCmmYPy7 BEGIN:VALARM ACTION:display DESCRIPTION:Deciphering Driving Behaviour using Geospatial Temporal Data C ollected from Smartphone Sensors in Auditorium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Lunch DTSTART:20160728T080000Z DTEND:20160728T091000Z DTSTAMP:20260421T155209Z UID:session/7G3EKTSWpH9bVE3V7u18Sm@hasgeek.com SEQUENCE:0 CREATED:20160606T051113Z DESCRIPTION:\n LAST-MODIFIED:20160720T090509Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Lunch in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:RightFit- A Data Science Approach to Reduce Product Returns in Fas hion e-Commerce DTSTART:20160728T091000Z DTEND:20160728T093000Z DTSTAMP:20260421T155209Z UID:session/GVtJyN6ZPjFsBkeX2xTq2K@hasgeek.com SEQUENCE:2 CATEGORIES:Crisp talk,Intermediate CREATED:20160719T100316Z DESCRIPTION:Refer to the attached presentation deck.\nMotivation\nProblem definition\nOur approach\nEvaluation\nConclusion\n\n### Speaker bio\n\nAsh ish Kulkarni works as a Principal data scientist at Jabong Labs. He's curr ently also a Ph.D. candidate in the Computer Science department at IIT Bom bay. His research interest is in the area of interactive machine learning and its application to information extraction and retrieval. Practical app lications like information extraction\, retrieval\, machine translation\, to name a few\, might benefit from autonomous machine learning models\, ai ded by user preferences. Interactive machine learning opens up promising a venues while posing research challenges in designing appropriate tools and algorithms. Studying and addressing these challenges forms the primary fo cus of his research. He has published in reputed conferences including IJC AI\, PAKDD\, K-CAP and others. Ashish has over eight years of prior indust ry experience.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/rightfit-a-data-scienc e-approach-to-reduce-product-returns-in-fashion-e-commerce-GVtJyN6ZPjFsBke X2xTq2K BEGIN:VALARM ACTION:display DESCRIPTION:RightFit- A Data Science Approach to Reduce Product Returns in Fashion e-Commerce in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Data-Driven Decision Making in Indian Agriculture: the Present and the Future DTSTART:20160728T091000Z DTEND:20160728T093000Z DTSTAMP:20260421T155209Z UID:session/GyiaTVP5FACH4mCfQMYZUz@hasgeek.com SEQUENCE:2 CATEGORIES:Crisp talk,Intermediate CREATED:20160720T090338Z DESCRIPTION:1. Introduction to data-driven decision making in agriculture. \n a. Overview of current decision-making process in agriculture sector.\n b. Discussion of the types of decisions that can be driven by data in the agriculture sector. \n2. The current state of agricul ture data in India. \n a. Overview of data that is openly available in ag riculture sector.\n b. Importance of specific data sets for in formation about agriculture in India.\n c. Insights from Gov ernment MIS systems\n3. Challenges and solutions in consolidating data.\n a. Overview of challenges faced in consolidating independent data se ts into a master data set that can be used for generating insights. \n b. Explanation of the solutions developed by us to stitch different datasets together.\n c. Data curation problems with the curren t agricultural data in India\n4. Regularity of data and data gaps i n the agricultural sector. \n a. Overview of current data gaps in agricul tural sector in India.\n b. Explanation of difference between input and o utput metrics in the context of agriculture. \n c. Argument for impo rtance of tracking output metrics at the lowest administrative di vision.\n5. Introduction to precision farming\n a. Overview of the concep t of precision farming. \n b. Explanation of how can sensor-based precisi on farming revolutionize the agriculture sector in India.\n c. Challenges to implementation of precision farming in the Ind ian context. \n6. Machine learning in agriculture\n a. How can machine le arning help in farmer preparedness on various issues? Predictin g crop productivity\, disease outbreaks\, pest infestations e tc. using machine learning algorithms.\n\n### Speaker bio\n\nUd it is a Data Scientist at SocialCops\, a data intelligence company in New Delhi. He has considerable experience in decision science and works on ana lyses to solve critical problems in agriculture\, health\, and retail. Pre vious projects include targeting agriculture investments to the people and places with the greatest impact\, analyzing the socio-economic potential of international markets\, and analyzing granular data on online retailers ' customer base. His mission is to create data-centered solutions to probl ems faced by agricultural India. Previous speaking experience includes deb ating and conducting sessions on the R programming language.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/data-driven-decision-m aking-in-indian-agriculture-the-present-and-the-future-GyiaTVP5FACH4mCfQMY ZUz BEGIN:VALARM ACTION:display DESCRIPTION:Data-Driven Decision Making in Indian Agriculture: the Present and the Future in Auditorium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:What do machine learning and high performance computing have to do with big cats in the wild? DTSTART:20160728T093000Z DTEND:20160728T103000Z DTSTAMP:20260421T155209Z UID:session/LXtogZMP8VMiv8sjKQjzf9@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160719T090827Z DESCRIPTION:In this talk\, focusing on some of my recent ecological resear ch work on tigers\, lions and cheetahs\, and quantitative methods\, I will describe what it takes to make computing approaches (ML\, PC and HPC) hig hly relevant to the field of ecology. I argue\, with a potpourri of demons trable examples\, that such computational “tools” are extremely useful only when considered within a broader scientific study and a failure to d o so may lead us into spurious conclusions that may prove costly!\n\n### S peaker bio\n\nDr. Arjun M. Gopalaswamy is visiting scientist at the Indian Statistical Institute and Research Associate at the Department of Zoology \, University of Oxford.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/what-do-machine-learni ng-and-high-performance-computing-have-to-do-with-big-cats-in-the-wild-LXt ogZMP8VMiv8sjKQjzf9 BEGIN:VALARM ACTION:display DESCRIPTION:What do machine learning and high performance computing have t o do with big cats in the wild? in Auditorium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Birds of Feather (BoF): How much math and statistics programmers n eed to know to hack their way into Machine Learning? DTSTART:20160728T093000Z DTEND:20160728T103000Z DTSTAMP:20260421T155209Z UID:session/9owrcBn7pnRnoUxMZTunLP@hasgeek.com SEQUENCE:1 CREATED:20160719T101036Z DESCRIPTION:An unconference session to discuss questions and ideas about h ow much math and statistics programmers need to know to hack their way int o machine learning.\n LAST-MODIFIED:20230108T103046Z LOCATION:Auditorium 3 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Birds of Feather (BoF): How much math and statistics programme rs need to know to hack their way into Machine Learning? in Auditorium 3 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Scaling the Largest Functional DataSet @Flipkart aka Catalog DTSTART:20160728T093000Z DTEND:20160728T101500Z DTSTAMP:20260421T155209Z UID:session/Bu9LkhYLVC5u9JRdS8Rpxj@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160719T100340Z DESCRIPTION:This required a shift in paradigm from traditional architectur es: replacing hardware LB with smart client\, emerging patterns like CQRS and focusing on techniques of optimization vertically as well as scaling h orizontally\, as we built this platform. I will share the process and lear nings in this talk.\n\n### Speaker bio\n\nAnuj is a SDE3 at flipkart. Curr ently working in cms team\, which is evolving catalog systems to store and serve high velocity semi-structured and unstructured catalog data. Prior to cms team\, he has worked as part of digital team\, platform team and a short stint in ELB team\, all at flipkart.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/scaling-the-largest-fu nctional-dataset-flipkart-aka-catalog-Bu9LkhYLVC5u9JRdS8Rpxj BEGIN:VALARM ACTION:display DESCRIPTION:Scaling the Largest Functional DataSet @Flipkart aka Catalog i n Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Increasing Trust and Efficiency of Data Science using dataset vers ioning DTSTART:20160728T101500Z DTEND:20160728T103500Z DTSTAMP:20260421T155209Z UID:session/3AJxoNznywjttazd691kCS@hasgeek.com SEQUENCE:2 CATEGORIES:Crisp talk,Intermediate CREATED:20160715T121700Z DESCRIPTION:1. Current process is iterative\, expensive\, and error prone\ n * Does not account for imperfectness in knowledge about the problem\, process\, organization\n * 80% of companies report strategic decisions going wrong due to flawed data \n2. Basic requirements of improved proces s - trust and efficiency\n * Trust requires auditability and reproducib ility of results \n * Efficiency requires standardization and automatio n \n3. Dataset is a fundamental abstraction of data science \n * Every data science task creates\, transforms\, validates\, and applies datasets\ n * Nesting and branching semantics \n4. New process around versioned d atasets \n * Import ideas from software engineering - versioning\, CI\, testing\n * Git & Github-like experience for datasets \n5. dgit - enab les git-like management of datasets \n * Python package\, open source\, MIT licence \n * Uses git for versioning\n * Focuses on capabilitie s that are specific to dataset management \n * Metadata management\ n * Inter-dataset dependency tracking\n * Scanning for datas et updates\n * Validation and generation\n * Support for met adata backends \n6. dgit implementation and demo\n * Architecture and f lexibility\n * Demos\n * Simplicity (automation)\n * Time line (lineage)\n * Validation of data and model results (trust\, au tomation)\n\n### Speaker bio\n\nDr. Venkata Pingali is Founder of Scribble Data\, a data science automation company. He was former VP\, Analytics at FourthLion technologies and led analytics work for large political campai gns and business customers of FourthLion. Previous to that he was Founder and CEO of an energy analytics company\, eLuminos. He has a BTech from IIT Mumbai and PhD from University of Southern California\, Los Angeles in sy stems\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/increasing-trust-and-e fficiency-of-data-science-using-dataset-versioning-3AJxoNznywjttazd691kCS BEGIN:VALARM ACTION:display DESCRIPTION:Increasing Trust and Efficiency of Data Science using dataset versioning in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Birds of Feather (BoF): Alternative Data DTSTART:20160728T103000Z DTEND:20160728T103500Z DTSTAMP:20260421T155209Z UID:session/EAHnM6J4Bh7ZnLKAXUEJ9p@hasgeek.com SEQUENCE:1 CREATED:20160719T100648Z DESCRIPTION:This is an unconference session to discuss questions and ideas about alternative / primary data and how to work with it.\n LAST-MODIFIED:20230108T103046Z LOCATION:Auditorium 3 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Birds of Feather (BoF): Alternative Data in Auditorium 3 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Evening tea break DTSTART:20160728T103500Z DTEND:20160728T111000Z DTSTAMP:20260421T155209Z UID:session/BnVYQdqTJp2yerg8eMhFHF@hasgeek.com SEQUENCE:0 CREATED:20160606T051541Z DESCRIPTION:\n LAST-MODIFIED:20160720T090518Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Evening tea break in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Hierarchical Structure\, Hierarchical Bayes approach and implemen tation of MCMC DTSTART:20160728T111000Z DTEND:20160728T115500Z DTSTAMP:20260421T155209Z UID:session/XbzdjsCE3mKVWLRapS1Wg1@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Advanced CREATED:20160718T123209Z DESCRIPTION:1. The talk begins by stating a couple of problems – one in ecology\, other in epidemiology. The characteristics of the problems will be explained with the types of inferences we are interested in. I will als o show the unavailability of any closed form solutions for them.\n2. This will be followed by a brief introduction to Bayesian methods of analysis a nd by Bayesian approach of learning from data. I will explain how hierarch ical structure helps in modelling data. Next will be a description of a ge neric set of Markov Chain Monte Carlo algorithms that is frequently used t o fit the hierarchical models.\n3. Then I will go back to the problems des cribed at the beginning of the talk and use hierarchical Bayesian approach with MCMC to solve them. I will explain usefulness and some drawbacks of MCMC under Bayesian approach. The talk will end after a discussion on gene rality and robustness of Bayesian paradigm.\n\n##Popular Quotes\n* `Inside every nonBayesian there is a Bayesian struggling to get out.’ - Dennis V. Lindley\n* `The practising Bayesian is well advised to become friends with as many numerical analysts as possible.’ - James O. Berger\n\n### S peaker bio\n\nSoumen Dey is currently a research scholar in Indian Statist ical Institute\, Bangalore. He has about 4 years of experience in handling ecological data and building statistical models. His research interests i nclude model selection\, modelling of data from multiple sources\, Bayesia n statistics.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/hierarchical-structure -hierarchical-bayes-approach-and-implementation-of-mcmc-XbzdjsCE3mKVWLRapS 1Wg1 BEGIN:VALARM ACTION:display DESCRIPTION:Hierarchical Structure\, Hierarchical Bayes approach and impl ementation of MCMC in Auditorium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Taking Analytics Applications from Labs to the Real World: Transfe r Learning in Practice DTSTART:20160728T111000Z DTEND:20160728T115500Z DTSTAMP:20260421T155209Z UID:session/2jEgXmU7GywDqc3s5Sk6Nv@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160719T091400Z DESCRIPTION:In the first half of this talk\, I will provide a brief overvi ew of Transfer Learning techniques touching upon theory\, applications\, a nd systems. In the second half\, I will talk about a real-life example ho w Transfer Learning can be effectively used for a social media analytics p roduct going over resultant benefits.\n\n### Speaker bio\n\nShourya Roy is currently a Senior Scientist and Research Manager at Xerox Research\, Ind ia where he leads the “Text and Graph Analytics” group. In this role\, Shourya is leading a group of researchers working on large scale text and graph analytics problems in domains such as outsourcing and customer care \, healthcare and education. As a part of this role he also looks after re search and business opportunities in customer care domain for Xerox in Sou th-East Asia and Australia. Shourya’s research interest spans Text and D ata Mining\, Natural Language Processing\, Machine Learning\, and Human Co mputation. Over the years\, Shourya’s work has led to about 50 patent di sclosures and over 50 publications in premier journals and conferences suc h as ACL\, AAAI\, SIGKDD\, SIGMOD\, VLDB\, WWW. He has taken up different professional roles including program committee member in top ranked confer ences\, editor in reputed journals\, reviewer of journal and conference pa pers\, advisor to students etc. He has been associated with several worksh ops in renowned text\, data and web mining conferences – notably\, the s eries of “Noisy Text Analytics”(AND) workshops which he co-initiated i n 2007. This year he is co-organizing two workshops viz. Network Data Anal ytics (NDA) with SIGMOD 2016 and Health Data Management and Mining (HDMM) with ICDE 2016.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/taking-analytics-appli cations-from-labs-to-the-real-world-transfer-learning-in-practice-2jEgXmU7 GywDqc3s5Sk6Nv BEGIN:VALARM ACTION:display DESCRIPTION:Taking Analytics Applications from Labs to the Real World: Tra nsfer Learning in Practice in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Djembe Jam DTSTART:20160728T115500Z DTEND:20160728T125500Z DTSTAMP:20260421T155209Z UID:session/Rdq2yaLi28m5xU38h81WVh@hasgeek.com SEQUENCE:1 CREATED:20160726T091704Z DESCRIPTION:Pick up the drums and play along with Ashok and Amit.\n LAST-MODIFIED:20230108T103046Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Djembe Jam in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Check-in and breakfast DTSTART:20160729T030000Z DTEND:20160729T035000Z DTSTAMP:20260421T155209Z UID:session/K7ZTyoUAwtuWj64GSye3V5@hasgeek.com SEQUENCE:0 CREATED:20160606T052006Z DESCRIPTION:\n LAST-MODIFIED:20160606T052009Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Check-in and breakfast in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Summary of day 1 DTSTART:20160729T035000Z DTEND:20160729T040000Z DTSTAMP:20260421T155209Z UID:session/KLNsmNpT6VM2iH12PpQUUB@hasgeek.com SEQUENCE:0 CREATED:20160606T052019Z DESCRIPTION:\n LAST-MODIFIED:20160606T052021Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Summary of day 1 in Auditorium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Let your Big Data Processing take flight with Apache Falcon DTSTART:20160729T044500Z DTEND:20160729T051500Z DTSTAMP:20260421T155209Z UID:session/XzZb5wcRoyzBJ6NEmGanZZ@hasgeek.com SEQUENCE:2 CATEGORIES:Crisp talk,Beginner CREATED:20160719T091221Z DESCRIPTION:The talk will mainly focus on the following areas:\n1. Why did InMobi create Falcon and what are the features it offers\n2. Overview of Falcon Architecture\n3. How Apache Falcon has solved some big data process ing problems at InMobi\n\nDraft slides on slideshare\n\n### Speaker bio\n\ nPallavi is an Architect at InMobi. She has been working on big data techn ologies for nearly 4 years now. She has deep knowledge of the Hadoop ecosy stem\, especially\, YARN\, PIG\, Oozie\, HBase\, Hive and Storm. She is th e committer for Apache Falcon and actively contributes to Apache PIG. She has spoken at conferences such as Annual RFID Conference \, Information Ma nagement Technical Conference\, ApacheCon (Big Data) and Grace Hopper Conf erence.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/let-your-big-data-proc essing-take-flight-with-apache-falcon-XzZb5wcRoyzBJ6NEmGanZZ BEGIN:VALARM ACTION:display DESCRIPTION:Let your Big Data Processing take flight with Apache Falcon in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Meet the needs of content marketing with the power of NLP DTSTART:20160729T044500Z DTEND:20160729T053000Z DTSTAMP:20260421T155209Z UID:session/ECiYU7QD66K1bqeVTsjscM@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160715T115819Z DESCRIPTION:I will start the talk by walking through what is Content Marke ting and its needs. I will layout the different stages of content marketin g. I will also outline some key challenges in Content Marketing. I will th en talk about a few techniques that can aid in Content Optimization. Here I will cover out-of-the-box techniques that can be explored for quick cont ent optimization such as summarizing a text as well as some advanced work in the direction extending the topics around a content. I will also talk a bout some recent works in the space of automatic content personalization f or specific demographies.\n\n### Speaker bio\n\nBalaji Vasan Srinivasan is a Computer Scientist at the Adobe Research Big data Experience Labs\, Ban galore\, India. His research interests span the areas of data mining\, nat ural language processing\, machine learning\, social data analytics and hi gh performance computing. He finished his Ph.D. in Computer Science at the University of Maryland in September 2011\, his thesis was on Scalable Lea rning Methods for Speaker Recognition and Geostatistics. Prior to that\, h e completed his M.S. in Electrical Engineering from University of Maryland in 2008 and B.E. in Electrical Engineering from Anna University (India) i n 2006. His research experience also includes stints at National Institute s of Health\, Bethesda\, MD (May – Aug 2007) and Xerox Research Center\, Webster\, NY (May – Aug 2011).\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/meet-the-needs-of-cont ent-marketing-with-the-power-of-nlp-ECiYU7QD66K1bqeVTsjscM BEGIN:VALARM ACTION:display DESCRIPTION:Meet the needs of content marketing with the power of NLP in A uditorium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Timely Dataflow DTSTART:20160729T051500Z DTEND:20160729T053500Z DTSTAMP:20260421T155209Z UID:session/8tLa71S2EF4XDxKrVWbpSh@hasgeek.com SEQUENCE:2 CATEGORIES:Crisp talk,Advanced CREATED:20160719T091304Z DESCRIPTION:What are the challenges faced in steram processing: Imagine a system where the data is continuously updated and you need to support both historical data + recent stream and avoid the costly recomputation \n\nHo w does timely dataflow fit in the stream processing model: Will be coverin g what timely dataflow offers - cyclic computation\, notification mechanis m\, concept of time in stream processing\n\nWhy is it different from other stream processing systems like spark/storm/flink : Not all computation ca n be easily expressed in Directed Acyclic Graphs which most of the stream processing systems offers - one such example is cyclic computation which c an be elegantly modelled in timely dataflow\n\nPros & Cons: Will take a pr actical example of an aggregation and showcase pros & cons of the timely d ataflow model \, with code and time taken\n\n### Speaker bio\n\nI am a pas sionate developer and a speaker. I regularly speak in the monthly geeknigh t meetup in chennai and have spoken in GIDS 2014\,2015 both the years on d ealing with systems that handle large volume of data with unique challenge s of near real time processing. I have built and maintained systems for Ba nking\, Media\, and Retail domain. I continuously challenge the status quo and constantly thrive to improve on the solutions i have built in the pas t. This journey has made me build & rebuild real time analytics solutions that crunches large volume of data carefully balancing throughput & low la tency\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/timely-dataflow-8tLa71 S2EF4XDxKrVWbpSh BEGIN:VALARM ACTION:display DESCRIPTION:Timely Dataflow in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Morning tea break DTSTART:20160729T053500Z DTEND:20160729T060500Z DTSTAMP:20260421T155209Z UID:session/N28uthJCVYM4eU349dhXHz@hasgeek.com SEQUENCE:0 CREATED:20160606T052148Z DESCRIPTION:\n LAST-MODIFIED:20160719T091413Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Morning tea break in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Keynote: Reasoning – The Next Frontier in Data Science DTSTART:20160729T060500Z DTEND:20160729T070500Z DTSTAMP:20260421T155209Z UID:session/VBZFUeaYSF7hR4PMZom269@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160722T075901Z DESCRIPTION:In this talk we will explore the next paradigm in data science - the “Reasoning Paradigm” that tries to optimize a “sequence of ac tions” leading from a “start state” to an “end state”. Prescribi ng a treatment plan for a set of symptoms\, learning strategies for playin g Chess or Go\, solving multi-step problems in mathematics\, maximizing li fe-time-value of a customer\, having a goal driven conversation with a cha t-bot\, or connecting the dots on a knowledge graph are different flavours of multi-step reasoning problems that cannot be solved by the single-step prediction paradigm. \n\nThis talk will focus on two specific reasoning paradigms - Mathematical Reasoning and Reasoning over Knowledge Graphs. We will explore the building blocks for an intelligent reasoning engine that “explores” the space of possible solutions\, “discovers” one or m ore solutions\, characterizes the "quality" of each solution\, “generali zes” to “similar” reasoning problems\, and most importantly “learn s” how to generate “better” solutions “faster” with practice - t he holy grail of AI.\n\n### Speaker bio\n\nShailesh Kumar is Chief Scienti st and Co-Founder at ThirdLeap. He has 14 years over fifteen years of expe rience in applying and innovating machine learning\, statistical pattern r ecognition\, and data mining algorithms to hard prediction problems in a w ide variety of domains including: remote sensing\, text mining\, bio-infor matics\, computer vision and image understanding\, transaction data mining \, retail analytics\, neurological data\, risk analytics in financial doma in\,and web analytics.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/reasoning-the-next-fro ntier-in-data-science-VBZFUeaYSF7hR4PMZom269 BEGIN:VALARM ACTION:display DESCRIPTION:Keynote: Reasoning – The Next Frontier in Data Science in A uditorium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Looking under the hood - demystifying data tools DTSTART:20160729T070500Z DTEND:20160729T072500Z DTSTAMP:20260421T155209Z UID:session/Psb3jDk6oxhWsPjNc2zF9d@hasgeek.com SEQUENCE:2 CATEGORIES:Crisp talk,Intermediate CREATED:20160715T120611Z DESCRIPTION:We will cover the design and development of experiments and pr esent benchmark results across select tabular (eg.: join\, aggregation etc .) and non-tabular operations (e.g. matrix multiplication\, sort/search et c.). For further analysis the code will be open-sourced soon after the ta lk.\n\n### Speaker bio\n\nSimrat is a Data Scientist\, Engineering Ninja a nd Inspector Gadget at Mad Street Den. She builds data platforms and model s to make sense of user and product data in e-commerce online retail.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/looking-under-the-hood -demystifying-data-tools-Psb3jDk6oxhWsPjNc2zF9d BEGIN:VALARM ACTION:display DESCRIPTION:Looking under the hood - demystifying data tools in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Lessons Learned : Real-life NLP DTSTART:20160729T070500Z DTEND:20160729T072500Z DTSTAMP:20260421T155209Z UID:session/3WM7CQ59rYSZS55CqzGmN2@hasgeek.com SEQUENCE:2 CATEGORIES:Crisp talk,Intermediate CREATED:20160720T090812Z DESCRIPTION:* Description of the real-life problem\, and the 'theoretical approach'\n* Theory vs Practice - when the 80/20 rule doesn't work\n* G etting a feel for Machine Learning in practice (and saving the day?)\n\n## # Speaker bio\n\nMartin has a PhD in Machine Learning\, and has been an Op en Source developer since 1999. After a career in finance (based in Londo n and New York)\, he decided to follow his original passion\, and now work s on Machine Learning / Artificial Intelligence full-time.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/lessons-learned-real-l ife-nlp-3WM7CQ59rYSZS55CqzGmN2 BEGIN:VALARM ACTION:display DESCRIPTION:Lessons Learned : Real-life NLP in Auditorium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Lunch DTSTART:20160729T072500Z DTEND:20160729T084000Z DTSTAMP:20260421T155209Z UID:session/6aZ5mEnkCDj5ozCKs8D9V6@hasgeek.com SEQUENCE:0 CREATED:20160606T052448Z DESCRIPTION:\n LAST-MODIFIED:20160729T050655Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Lunch in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Scalable Realtime Analytics using Druid DTSTART:20160729T084000Z DTEND:20160729T092500Z DTSTAMP:20260421T155209Z UID:session/6xMUtccL3omAhwcZPCBUwZ@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160719T091205Z DESCRIPTION:- History and Motivation \n- Live Demo\n- Druid Architecture\n - Storage Internals \n- Druid in Practice\n- Common Use Cases\n\n### Speak er bio\n\nNishant is an active contributor and PMC member for Druid. He is part of Business Intelligence team at Hortonworks. Prior to that he was p art of Metamarkets backend team and was responsible for analytics infrastr ucture\, including real-time analytics in Druid. He holds a B.Tech in Comp uter Science from National Institute of Technology\, Kurukshetra\, India.\ n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/scalable-realtime-anal ytics-using-druid-6xMUtccL3omAhwcZPCBUwZ BEGIN:VALARM ACTION:display DESCRIPTION:Scalable Realtime Analytics using Druid in Auditorium 2 in 5 m inutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Continuous online learning for classification tasks DTSTART:20160729T084000Z DTEND:20160729T092500Z DTSTAMP:20260421T155209Z UID:session/KGL6262Q3LazFVYwKKZJRb@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160715T115310Z DESCRIPTION:* Introduction to the problem of online continuous learning\n* Examples of Concept drift\n* Capturing of concept drift using perceptron\ n* Fallacies of a local-global model and the need for ensembling\n* Toward s a robust ensembling manager\n\n### Speaker bio\n\nI am an entrepreneur a nd machine learning researcher. I dropped out of my doctoral program in 20 12 and founded Airwoot\, a company that help businesses deliver customer s upport on social using lot of mathematical tricks. Airwoot is now acquired by Freshdesk where I continue to build the technology that can teach mach ines to learn about natural language and emotions. \n\nI completed my mast ers from DTU Denmark and was pursuing Ph.D from Hasso-plattner Institute\, Berlin. In my academic career I contributed research at Macquarie Univers ity\, Infocomm Research Singapore and the Swedish Institute of Computer Sc ience.\n\nhttps://in.linkedin.com/in/saurabhairwoot\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/continuous-online-lear ning-for-classification-tasks-KGL6262Q3LazFVYwKKZJRb BEGIN:VALARM ACTION:display DESCRIPTION:Continuous online learning for classification tasks in Auditor ium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Convolutional Neural Networks from the Other Side DTSTART:20160729T092500Z DTEND:20160729T101000Z DTSTAMP:20260421T155209Z UID:session/3SdH19f1smTCvYx92EQnxS@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Advanced CREATED:20160715T120536Z DESCRIPTION:Outline: \n* What is new about Deep Learning and what were the paradigm changes\n* Interesting results in CNN Architectures: GoogLeNet\, Residual Nets\, Sim Nets\, Spatial Transformer Networks\n* Results in Vis ualization and Debugging\n* Newer update Mechanisms and Optimization Resul ts\n* Theoretical Results (Current interpretation on why these models work ): Tensor Decomposition perspective\, Stephane Mallat and Joan Bruna work (If time permits)\n\n### Speaker bio\n\nSumod Mohan holds an M.S degree in Electrical Engineering from Clemson University and led the Computer Visio n efforts at HighlightCam (a Bay Area Startup) before joining Soliton. His experience spans Computer Vision\, Machine Learning\, 3D Vision\, Deep Le arning\, Graph Algorithms\, Probabilistic Graphical Models\, Code Optimiza tion and Parallelization and has worked in the Computer Vision and Machine Learning for past 10 years. He currently leads the Computer Vision and Ma chine Learning Business Division (CVD) at Soliton Technologies and was the speaker at Fifth Elephant 2015.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/convolutional-neural-n etworks-from-the-other-side-3SdH19f1smTCvYx92EQnxS BEGIN:VALARM ACTION:display DESCRIPTION:Convolutional Neural Networks from the Other Side in Auditoriu m 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Dr. Elephant - Self-Serve Performance Tuning for Hadoop and Spark DTSTART:20160729T092500Z DTEND:20160729T095500Z DTSTAMP:20260421T155209Z UID:session/EgTyhQKpZYmhK3YKn9rHvt@hasgeek.com SEQUENCE:2 CATEGORIES:Crisp talk,Intermediate CREATED:20160715T121518Z DESCRIPTION:Phase 1: I'll share the experience at Linkedin in optimizing t he user jobs\, the challenges we faced and how a simple self serve tool li ke Dr. Elephant helped overcome these challenges. \n\nPhase 2: I’ll shar e how we integrated such a tool into our developer lifecycle and encourage d them to optimize the jobs with minimal support from the hadoop experts. \n\nPhase 3: This phase will involve discussions about the tool\, how it a nalyses the job by gathering all the diverse information\, how to write cu stom heuristics and plug them into Dr. Elephant\, comparing and analysing job executions etc.\n\n### Speaker bio\n\nAkshay Rai is an engineer at Lin kedin working for the Hadoop development team. He has been working on Dr. Elephant for more than a year and has worked extensively to help open sour ce this tool. Since the open source announcement last week\, he has been a ctively engaging in discussions with the community and leading this projec t.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/dr-elephant-self-serve -performance-tuning-for-hadoop-and-spark-EgTyhQKpZYmhK3YKn9rHvt BEGIN:VALARM ACTION:display DESCRIPTION:Dr. Elephant - Self-Serve Performance Tuning for Hadoop and Sp ark in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Evening tea break DTSTART:20160729T101000Z DTEND:20160729T105000Z DTSTAMP:20260421T155209Z UID:session/DmHb7TBSw5oJJUu1PE5uJz@hasgeek.com SEQUENCE:0 CREATED:20160606T052740Z DESCRIPTION:\n LAST-MODIFIED:20160729T054636Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Evening tea break in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Hadoop & Cloud Storage: Object Store Integration in Production DTSTART:20160729T105000Z DTEND:20160729T111500Z DTSTAMP:20260421T155209Z UID:session/QrpZdgniM9QyTV2BfQSnZS@hasgeek.com SEQUENCE:2 CATEGORIES:Crisp talk,Intermediate CREATED:20160715T121442Z DESCRIPTION:In this session\, I am going to explore challenges mentioned i n abstract and present recent work to address them in a comprehensive effo rt spanning multiple Hadoop ecosystem components\, including the Object St ore FileSystem connector\, Hive\, Tez and ORC. Our goal is to improve corr ectness\, performance\, security and operations for users that choose to i ntegrate Hadoop with Cloud Storage. We use S3 and S3A connector as case st udy.\n\n### Speaker bio\n\nRajesh Balamohan is a "Member of Technical Staf f" in Hortonworks. He has been working on Hadoop for last couple of years. Recently he has been concentrating on Tez performance at scale. Rajesh is a committer and PMC in Apache Tez project.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/hadoop-cloud-storage-o bject-store-integration-in-production-QrpZdgniM9QyTV2BfQSnZS BEGIN:VALARM ACTION:display DESCRIPTION:Hadoop & Cloud Storage: Object Store Integration in Production in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Birds of Feather (BoF): From Machine Learning to Deep Learning –  when\, how and why do you make the transition? DTSTART:20160729T105000Z DTEND:20160729T112500Z DTSTAMP:20260421T155209Z UID:session/Hyi15SS1qxG6A2uoEAsuG3@hasgeek.com SEQUENCE:1 CREATED:20160719T101401Z DESCRIPTION:This is an unconference session on when\, how and why should y ou make the transition from machine learning to deep learning.\n LAST-MODIFIED:20230108T103046Z LOCATION:Auditorium 3 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Birds of Feather (BoF): From Machine Learning to Deep Learning – when\, how and why do you make the transition? in Auditorium 3 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Model Visualisation DTSTART:20160729T105000Z DTEND:20160729T113500Z DTSTAMP:20260421T155209Z UID:session/FbFY9ZqrpT8vUUGYEHUsRc@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Beginner CREATED:20160719T091741Z DESCRIPTION:Data science is a process of abstraction. In order to explain or to predict a real phenomena\, the process start with framing the proble m\, acquiring & refining the data and then moves between the three layers of abstraction - transformations (data abstraction)\, visualisations (visu al abstraction) and modelling (symbolic abstraction). All these three laye rs of abstraction work together to try and build a truer (or more closer) representation of the real phenomena.\n\n**Data visualisation (data-vis)** helps us to understand the portrait and the shape of the data. The scienc e of data-vis for exploratory data analysis is well developed\, for both s tatic graphics (scatter plot matrices\, glyph based approaches\, geometric transforms like parallel coordinates) and interactive graphics (layering\ , brushing and linking\, projections and tours). See my talk at Fifth Elep hant 2015 on Visualising Multi Dimensional Data - [https://www.youtube.com /watch?v=X8rNDvPNg30](https://www.youtube.com/watch?v=X8rNDvPNg30). Howeve r\, the power of visualisation is rarely leveraged for understanding the m odels developed better. Model evaluation is still largely restricted throu gh numerical summaries. Extending visualisation to model building can be a powerful way to improve our understanding of the model.\n\n**Model visual isation (model-vis)** can help us to understand the shape of the model and compare it to the shape of the data. It allows to see the fit of the mode l and understand where the fit can be improved. It also allows us to bette r understand the parameters in the model and how the model changes when th e parameters change as well as how the parameters changes when the input d ata changes. \n\nThe science and tools for model-vis are still very under- developed. This talks looks at practical examples of doing model-vis in re gression (linear\, lasso)\, classification (logistic\, trees\, LDA) and cl ustering (hierarchical) problems that can help us better understand the mo del. This includes exploring model-vis approaches that:\n\n- Visualise th e model in data space as opposed to data in model space\n- Visualise the entire space of models \n- Visualise the same model with varying tuning p arameters\n- Visualise the same model with different input datasets\n- V isualise the process of model fitting as opposed to final result\n\nIntegr ating these approaches for model-vis as a part of model evaluation will st rengthen the understanding of the model and lead to better model building for a data scientist. Model-vis can then complement data-vis for fitting b etter models as well as for communicating the insight from the data scienc e process.\n\nPost this talk\, the audience will have a better understandi ng of the power of visualisation beyond data-vis to model-vis and use it t o build better models as a data scientist.\n\n### Speaker bio\n\nAmit Kapo or is interested in learning and teaching the craft of telling visual stor ies with data. He uses storytelling and data visualization as tools for im proving communication\, persuasion and leadership. He conducts workshops a nd trainings on data visualisation and data science for corporates\, non-p rofits\, colleges\, and individuals at narrativeVIZ Consulting. He also te aches storytelling with data as invited faculty in management schools e.g. IIM Bangalore\, IIM Ahmedabad and design schools e.g. NID Bangalore.\n\nH is background is in strategy consulting in using data-driven stories to dr ive change across organizations and businesses. He has more than 14 years of management consulting experience\, first with AT Kearney in India\, the n with Booz & Company in Europe and more recently for startups in Bangalor e. He did his B.Tech in Mechanical Engineering from IIT\, Delhi and PGDM ( MBA) from IIM\, Ahmedabad. You can find more about him at [amitkaps.com](h ttp://amitkaps.com) and tweet him at [@amitkaps](http://twitter.com/amitka ps).\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/model-visualisation-Fb FY9ZqrpT8vUUGYEHUsRc BEGIN:VALARM ACTION:display DESCRIPTION:Model Visualisation in Auditorium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Machine Learning the Walmart Way with a Deep Dive into Automated F orecasting System DTSTART:20160729T111500Z DTEND:20160729T114000Z DTSTAMP:20260421T155209Z UID:session/JzUgWrf8KGhcNahexm6mLT@hasgeek.com SEQUENCE:2 CATEGORIES:Crisp talk,Intermediate CREATED:20160719T092200Z DESCRIPTION:In this talk we are going to talk about how we go about implem enting machine learning at Walmart with deep dive into our forecasting arc hitecture.\n\n### Speaker bio\n\nAnindya Sankar Dey is currently an Associ ate Data Scientist in Walmart Technology. He has about 7 years of experien ce in data science and machine learning domain. He completed his Masters i n Statistics from ISI Kolkata.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/machine-learning-the-w almart-way-with-a-deep-dive-into-automated-forecasting-system-JzUgWrf8KGhc Nahexm6mLT BEGIN:VALARM ACTION:display DESCRIPTION:Machine Learning the Walmart Way with a Deep Dive into Automat ed Forecasting System in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Real Time Fulfilment Planning at Flipkart Scale DTSTART:20160729T113500Z DTEND:20160729T122000Z DTSTAMP:20260421T155209Z UID:session/JxCT1WYQiPxzHXav1wmC88@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160719T092112Z DESCRIPTION:In this talk I am presenting how planning is done in real time to ensure that SLAs given to customers are met considering the capacity a nd constraints of the supply chain network and at lowest cost.\n\n### Spea ker bio\n\nJagadeesh is an Architect at Flipkart. He is currently working on Flipkart's Supply Chain and Logistics Platforms focussing on Fulfilment Planning and Orchestration. Prior to this he was architect for Payments\, Settlement and Fraud Detection Systems at Flipkart. He has built and oper ated several scaleable and highly available distributed systems. His inter ests include Change Propagation\, Messaging Systems\, Databases and Operat ions Research that focusses on Logistics and Routing problems.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/real-time-fulfilment-p lanning-at-flipkart-scale-JxCT1WYQiPxzHXav1wmC88 BEGIN:VALARM ACTION:display DESCRIPTION:Real Time Fulfilment Planning at Flipkart Scale in Auditorium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Logging at scale using Graylog - Billion+ messages\, 100K req/sec DTSTART:20160729T114000Z DTEND:20160729T120000Z DTSTAMP:20260421T155209Z UID:session/D9FqE3Gc1kHk7bR8z4vFUy@hasgeek.com SEQUENCE:2 CATEGORIES:Crisp talk,Intermediate CREATED:20160720T090850Z DESCRIPTION:Talk Outline:\n\n* Infrastructure Overview:\n * Hundreds of micro-services\n * 100k requests per second\n* The logging pipeline: K afka\, Graylog & Elasticsearch\n* Scalability Issues\, Resolution and less ons learnt:\n * Huge Lag for Application logs in Graylog UI\n * Dock er service crashing due to Fluentd log driver\n * Exceptions in Graylog server due to 3MB log messages\n * Journal Utilisation too high\, unco mmitted messages deleted from journal - Part I\n * Journal Utilisation too high\, uncommitted messages deleted from journal - Part II\n * Slow Output compared to Inputs from Kafka\n\n### Speaker bio\n\nRohit is a tec hnologist\, explorer and a proud Indian.\nAs a technologist\, he has worke d in the area of convergence of telephony over the web. He has worked as a n infrastructure engineer\, product developer\, did customer support\, hel ped built teams and also led a team of 10 DevOps developers. He loves open -source\, started a Linux User Group at his university and has an active G ithub profile.\nAs an explorer\, he loves to try out new things. He loves trekking\, photography\, challenged himself with vipassana meditation and long distance bike tours.\n\nYou can find more about him at www.rohit.io\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/logging-at-scale-using -graylog-billion-messages-100k-req-sec-D9FqE3Gc1kHk7bR8z4vFUy BEGIN:VALARM ACTION:display DESCRIPTION:Logging at scale using Graylog - Billion+ messages\, 100K req/ sec in Auditorium 2 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Reducing the world with JavaScript DTSTART:20160729T120000Z DTEND:20160729T122500Z DTSTAMP:20260421T155209Z UID:session/QfVaYx5WAesfFgWLzpuyvN@hasgeek.com SEQUENCE:2 CATEGORIES:Full talk,Intermediate CREATED:20160719T091946Z DESCRIPTION:A single image of the world containing street level data would be far too large to be held in memory or to be able to be downloaded at o nce - which leads us to map tiles. Map tiles are usually 256px by 256px in size and are placed next to each other to create the illusion of a single image. \n\nTiles can be either rasters or vectors. While raster tiles con tain pixel information and require processing on the server side before be ing rendered\, vector tiles hold tile data in either a human readable _geo json_ format or as _protobufs_\, making them easier and faster to process and render on the client side. Vector tiles also contain useful geodata th at can be parsed.\n\n![tr](https://cloud.githubusercontent.com/assets/3166 852/14909657/cb28838a-0e03-11e6-89e0-b0ecb46aa66f.gif)\n\nGreater the numb er of tiles that a map is composed of\, the greater detail that map can sh ow. In order to manage millions of tile images/data\, web maps use a simpl e coordinate system. Each tile has a z coordinate describing its zoom leve l and x and y coordinates describing its position within a square grid for that zoom level: z/x/y.\n\nZoom levels are related to each other by power s of four:\n\n * z0 contains 1 tile.\n * z1 contains 4 tiles.\n * z2 co ntains 16 tiles.\n\n ....\n \n ....\n \n .... \n\n * zn contains 2^n * 2^n tiles\n\nAs you can see\, the number of tile s increases exponentially with the zoom level\, which leads to an exponent ial increase in bandwidth and memory requirements\, not to mention a great er difficulty in parsing and analysing such a lot of data.\n\nTileReduce w as written to process\, or mine geodata from\, these millions of tiles asy nchronously using the MapReduce concept on vector tiles\, making it one of the fastest ways to parse tile data for the whole world. Compared to the complex postgres queries you would have to write to do the same operations \, TileReduce is also an extremely easy library to use.\n\n**Outline** \n\ n* Web Maps\n* Tiles\n * Vector and Raster\n * Tile formats. \n * What does a simple\, human readable vector tile contain?\n* JavaScript Ma p Reduce\n* Asynchronous processing\n* Vector Tiles and Map Reduce\n* Tile Reduce \n * Background and history.\n * Vector tiles - tools to conv ert certain data types to vector tiles.\n * Program skeleton\n * Exa mples\n * Limitations\n* Q&A\n\n### Speaker bio\n\nI am a developer at Mapbox. I've built several programs that use TileReduce to run analyses on OpenStreetMap data\, and have been part of sessions at Mapbox on the JS p rinciples that TileReduce was built on - so I understand the theory well. I also think it is one of the most beautiful tools to process geographic d ata and hope to share this same fascination that I have for the tool with the audience at the Fifth Elephant. Earlier\, I have written 2D games usin g the Cocos2D engine\, submitted small activities to the GCompris project and written application documentation for the GNOME Foundation. I also enj oy collecting recordings of Indian classical music that are in the public domain.\n LAST-MODIFIED:20230810T072606Z LOCATION:Auditorium 2 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/reducing-the-world-wit h-javascript-QfVaYx5WAesfFgWLzpuyvN BEGIN:VALARM ACTION:display DESCRIPTION:Reducing the world with JavaScript in Auditorium 2 in 5 minute s TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Feedback and theme for 2017 DTSTART:20160729T122500Z DTEND:20160729T124500Z DTSTAMP:20260421T155209Z UID:session/JNCJogm3PPPwoEzm7uoKJx@hasgeek.com SEQUENCE:0 CREATED:20160726T092059Z DESCRIPTION:\n LAST-MODIFIED:20160729T050655Z LOCATION:Auditorium 1 - NIMHANS Convention Centre\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Feedback and theme for 2017 in Auditorium 1 in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Check-in and Breakfast DTSTART:20160730T031500Z DTEND:20160730T040000Z DTSTAMP:20260421T155209Z UID:session/Jm7xhm5QwL2bDBUFcVmzpa@hasgeek.com SEQUENCE:0 CREATED:20160623T155523Z DESCRIPTION:\n LAST-MODIFIED:20160706T085738Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Check-in and Breakfast in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Check-in and Breakfast DTSTART:20160730T031500Z DTEND:20160730T040000Z DTSTAMP:20260421T155209Z UID:session/EUqeMpGiFEVnTM2VJufprX@hasgeek.com SEQUENCE:0 CREATED:20160713T094912Z DESCRIPTION:\n LAST-MODIFIED:20160718T061719Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Check-in and Breakfast in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Advanced Deep Learning Workshop – Hands-on DTSTART:20160730T040000Z DTEND:20160730T053000Z DTSTAMP:20260421T155209Z UID:session/322uyDFif97qFaxyKSXp29@hasgeek.com SEQUENCE:2 CATEGORIES:Workshop,Advanced CREATED:20160713T094934Z DESCRIPTION:The workshop will start from the very basics (with a little ma thematics)\, and quickly progress to getting hands-on with open source sof tware including the training of a deep network on simple problems to get ' warmed up'.\n\nThis will be followed by several deeper dives using a pre-b uilt Virtual Machine\, running within VirtualBox. Participants will exper iment with a much larger pre-trained models\, and get an understanding of several application areas\, among which are : \n\n * Anomaly detection\n \n * Applying a pre-trained model to classify images into previously uns een classes\n\n * Art 'Style-Transfer'\n\n * Reinforcement Learning (i nspired by AlphaGo)\n\n\nWhile parts of this are very technical\, the mode ls (inside the Virtual Machine) are all in Jupyter (fka iPython) notebooks \, making interaction straightforward. \n\nThe Python libraries that are used are Theano and Lasagne (both on GitHub).\n\n### Speaker bio\n\nMartin has a PhD in Machine Learning\, and has been an Open Source developer sin ce 1999. After a career in finance (based in London and New York)\, he de cided to follow his original passion\, and now works on Machine Learning / Artificial Intelligence full-time.\n LAST-MODIFIED:20230810T072606Z LOCATION:Hall - TERI\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/advanced-deep-learning -workshop-hands-on-322uyDFif97qFaxyKSXp29 BEGIN:VALARM ACTION:display DESCRIPTION:Advanced Deep Learning Workshop – Hands-on in Hall in 5 min utes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Introduction to Statistics and Basics of Mathematics for Data Scie nce - the hacker's way DTSTART:20160730T040000Z DTEND:20160730T053000Z DTSTAMP:20260421T155209Z UID:session/Qq3kVJnDQpeLoZzt5igwVo@hasgeek.com SEQUENCE:2 CATEGORIES:Workshop,Beginner CREATED:20160623T155825Z DESCRIPTION:This is a full-day workshop. The 6-hour workshop is roughly sp lit into 3 major modules. Each module will introduce some math and then an application is introduced where the concepts learnt will be used. \n\n# W orkshop Topics and Structure\n\n**Module 1: Basics of Statistics (Applicat ion: A/B Testing)**\n\nThe first part of this module will introduce the ba sic concepts (mean\, median\, standard deviation\, variance\, probability distribution). Then\, using A/B testing as application\, hypothesis testin g is introduced. At the end of this module\, attendees will be able to und erstand what confidence intervals are\, significance levels\, confidence i ntervals\, p-value and t-test. \n\n**Module 2: Basics of Linear Algebra (Application: Supervised Machine Learning: Linear Regression)**\n\nThe fir st part of this module will introduce attendees to the world of linear alg ebra (vectors\, matrices and operations on them). One of the simplest and most powerful supervised machine learning algorithm\, linear regression\, is introduced using an application where the attendees are taught how to b uild a predictive model to predict a continuous target variable. The vario us diagnostics from the linear model's output are discussed. \n\n**Module 3: Basics of Linear Algebra -continued (Application: Unsupervised Machine Learning: Dimensionality Reduction)**\n\nIn the first part of this module\ , eigen value and eigen vectors are introduced. Then an unsupervised machi ne learning algorithm\, Principal Component Analysis\, is introduced and a n application of dimensionality reduction is implemented. \n\nDepending on time and interest\, one of the clustering algorithms - *k-means clusterin g algorithm* will be implemented. \n\n\n**Software Requirements for the Wo rkshop:**\n\nWe will be using *Python* data stack for the workshop. \n\nPl ease install Ananconda for Python 3.5 for the workshop. That has everythin g we need for the workshop. \n\nFor attendees more curious\, we will be us ing Jupyter Notebook as our IDE. We will be introducing `numpy`\, `scipy`\ , `seaborn`\, `matplotlib`\, `statsmodel` and `scikit-learn`. \n\n*No know ledge of Python is assumed*\n\n\n**Data Repository for the Workshop:**\n\n The data necessary for the workshop will be available in the github reposi tory two weeks before the workshop. Please download them before coming for the workshop. The repository for the workshop is:\nhttps://github.com/ami tkaps/hackermath\n\n### Speaker bio\n\nAmit Kapoor teaches the craft of te lling visual stories with data. He conducts workshops and trainings on Dat a Science in Python and R\, as well as on Data Visualisation topics. His b ackground is in strategy consulting having worked with AT Kearney in India \, then with Booz & Company in Europe and more recently for startups in Ba ngalore. He did his B.Tech in Mechanical Engineering from IIT\, Delhi and PGDM (MBA) from IIM\, Ahmedabad. You can find more about him at http://ami tkaps.com/ and tweet him at [@amitkaps](https://www.twitter.com/amitkaps). \n\nBargava Subramanian is a Data Scientist at Cisco Systems\, India. He h as 14 years of experience delivering business analytics solutions to Inves tment Banks\, Entertainment Studios and High-Tech companies. He has given talks and conducted workshops on Data Science\, Machine Learning\, Deep Le arning and Optimization in Python and R. He has a Masters in Statistics fr om University of Maryland\, College Park\, USA. He is an ardent NBA fan. Y ou can tweet to him at [@bargava](https://www.twitter.com/bargava).\n LAST-MODIFIED:20230810T072606Z LOCATION:C5 (Seminar Hall) - 91springboard\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/introduction-to-statis tics-and-basics-of-mathematics-for-data-science-the-hackers-way-Qq3kVJnDQp eLoZzt5igwVo BEGIN:VALARM ACTION:display DESCRIPTION:Introduction to Statistics and Basics of Mathematics for Data Science - the hacker's way in C5 (Seminar Hall) in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Tea Break DTSTART:20160730T053000Z DTEND:20160730T054500Z DTSTAMP:20260421T155209Z UID:session/E4cxN3DtaMbfBRJSQtKKkZ@hasgeek.com SEQUENCE:0 CREATED:20160623T155447Z DESCRIPTION:\n LAST-MODIFIED:20160623T155449Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Tea Break in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Tea Break DTSTART:20160730T053000Z DTEND:20160730T054500Z DTSTAMP:20260421T155209Z UID:session/GqLRiMyLKWfB1tM5uJL2uD@hasgeek.com SEQUENCE:0 CREATED:20160713T095011Z DESCRIPTION:\n LAST-MODIFIED:20160713T095016Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Tea Break in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Contd. Advanced Deep Learning Workshop – Hands-on DTSTART:20160730T054500Z DTEND:20160730T071500Z DTSTAMP:20260421T155209Z UID:session/9wTFX7UK9WXn7QXszoSEH@hasgeek.com SEQUENCE:0 CREATED:20160713T095028Z DESCRIPTION:\n LAST-MODIFIED:20160718T061741Z LOCATION:Hall - TERI\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Contd. Advanced Deep Learning Workshop – Hands-on in Hall i n 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Contd. Introduction to Statistics and Basics of Mathematics for Da ta Science - the hacker's way DTSTART:20160730T054500Z DTEND:20160730T071500Z DTSTAMP:20260421T155209Z UID:session/Kc8oeAzWYQvrRyXD4wPwHH@hasgeek.com SEQUENCE:0 CREATED:20160623T155953Z DESCRIPTION:\n LAST-MODIFIED:20160718T062023Z LOCATION:C5 (Seminar Hall) - 91springboard\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Contd. Introduction to Statistics and Basics of Mathematics fo r Data Science - the hacker's way in C5 (Seminar Hall) in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Lunch DTSTART:20160730T071500Z DTEND:20160730T081500Z DTSTAMP:20260421T155209Z UID:session/FbJsV8UbWno9rZJ6JzNRCM@hasgeek.com SEQUENCE:0 CREATED:20160713T095120Z DESCRIPTION:\n LAST-MODIFIED:20160713T095130Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Lunch in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Lunch DTSTART:20160730T071500Z DTEND:20160730T081500Z DTSTAMP:20260421T155209Z UID:session/PyxCi7iz3KaVN8ceLjfuJE@hasgeek.com SEQUENCE:0 CREATED:20160623T155542Z DESCRIPTION:\n LAST-MODIFIED:20160706T085907Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Lunch in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Contd. Introduction to Statistics and Basics of Mathematics for Da ta Science - the hacker's way DTSTART:20160730T081500Z DTEND:20160730T100000Z DTSTAMP:20260421T155209Z UID:session/EHtAhXY5y61pV6hePst6zv@hasgeek.com SEQUENCE:0 CREATED:20160623T160112Z DESCRIPTION:\n LAST-MODIFIED:20160718T062037Z LOCATION:C5 (Seminar Hall) - 91springboard\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Contd. Introduction to Statistics and Basics of Mathematics fo r Data Science - the hacker's way in C5 (Seminar Hall) in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Contd. Advanced Deep Learning Workshop – Hands-on DTSTART:20160730T081500Z DTEND:20160730T100000Z DTSTAMP:20260421T155209Z UID:session/XZyEsEqKoBS7hLk7yN1pN7@hasgeek.com SEQUENCE:0 CREATED:20160713T095137Z DESCRIPTION:\n LAST-MODIFIED:20160718T061812Z LOCATION:Hall - TERI\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Contd. Advanced Deep Learning Workshop – Hands-on in Hall i n 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Tea Break DTSTART:20160730T100000Z DTEND:20160730T101500Z DTSTAMP:20260421T155209Z UID:session/GSJQ6DpMuNVjh7jLSpqkiH@hasgeek.com SEQUENCE:0 CREATED:20160713T095202Z DESCRIPTION:\n LAST-MODIFIED:20160713T095207Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Tea Break in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Tea Break DTSTART:20160730T100000Z DTEND:20160730T101500Z DTSTAMP:20260421T155209Z UID:session/B4Rt42ehz1QkTjLWdwmALh@hasgeek.com SEQUENCE:0 CREATED:20160623T155630Z DESCRIPTION:\n LAST-MODIFIED:20160706T085916Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Tea Break in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Contd. Advanced Deep Learning Workshop – Hands-on DTSTART:20160730T101500Z DTEND:20160730T113000Z DTSTAMP:20260421T155209Z UID:session/5TTt6w3bEtkEDmCa3xeGr7@hasgeek.com SEQUENCE:0 CREATED:20160713T095216Z DESCRIPTION:\n LAST-MODIFIED:20160718T061824Z LOCATION:Hall - TERI\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Contd. Advanced Deep Learning Workshop – Hands-on in Hall i n 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Contd. Introduction to Statistics and Basics of Mathematics for Da ta Science - the hacker's way DTSTART:20160730T101500Z DTEND:20160730T113000Z DTSTAMP:20260421T155209Z UID:session/7rsEtjYS4pHk1Gr86Z1S51@hasgeek.com SEQUENCE:0 CREATED:20160623T160220Z DESCRIPTION:\n LAST-MODIFIED:20160718T062046Z LOCATION:C5 (Seminar Hall) - 91springboard\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Contd. Introduction to Statistics and Basics of Mathematics fo r Data Science - the hacker's way in C5 (Seminar Hall) in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Check-in and Breakfast DTSTART:20160731T024500Z DTEND:20160731T033000Z DTSTAMP:20260421T155209Z UID:session/RpLfNZbdkRNZPk7dcSEDRY@hasgeek.com SEQUENCE:0 CREATED:20160623T155846Z DESCRIPTION:\n LAST-MODIFIED:20160726T055317Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Check-in and Breakfast in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Deep Learning for Computer Vision DTSTART:20160731T033000Z DTEND:20160731T053000Z DTSTAMP:20260421T155209Z UID:session/JUpB5kDQpJR5e3sz5xw71v@hasgeek.com SEQUENCE:2 CATEGORIES:Workshop,Intermediate CREATED:20160623T155131Z DESCRIPTION:Some of the theory covered will be:\n\n1. From neurons to netw orks\, a full overview of the nuts and bolts.\n2. Types of networks\, from RBMs to CNNs to RNNs.\n3. How they are used in the world of CV. A discuss ion of what works and what doesn't.\n4. The fundamentals of training Deep Learning networks. \n5. On existing frameworks and using GPUs.\n\nA relati vely large number of potential projects will be available for the workshop \, and the participants will have access to AWS ec2 GPU instances for trai ning and testing.\nSome Potential workshop projects:\n\n1. Toy problems ex ploring the different architectures and relative merits.\n2. Real world cl assification problems solved with different architectures.\n3. Exploring n ovel and hybrid architectures (if time permits).\n\n### Speaker bio\n\nDr. Anand Chandrasekaran is a founder and the CTO of Mad Street Den\, an AI c ompany specializing in computer vision. In addition to an academic backgro und in the fields of neuroscience and neuromorphic engineering\, he has be en a member of teams working on DARPA projects in cognition and vision.\n LAST-MODIFIED:20230810T072606Z LOCATION:Hall - TERI\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/deep-learning-for-comp uter-vision-JUpB5kDQpJR5e3sz5xw71v BEGIN:VALARM ACTION:display DESCRIPTION:Deep Learning for Computer Vision in Hall in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Tea Break DTSTART:20160731T053000Z DTEND:20160731T054500Z DTSTAMP:20260421T155209Z UID:session/21Wdy13SxurhCM9e82pZZ6@hasgeek.com SEQUENCE:0 CREATED:20160623T155925Z DESCRIPTION:\n LAST-MODIFIED:20160623T160020Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Tea Break in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Contd. Deep Learning for Computer Vision DTSTART:20160731T054500Z DTEND:20160731T074500Z DTSTAMP:20260421T155209Z UID:session/JfGu9DiKEhGj2UwyXHQyF4@hasgeek.com SEQUENCE:0 CREATED:20160623T155256Z DESCRIPTION:\n LAST-MODIFIED:20160726T055236Z LOCATION:Hall - TERI\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Contd. Deep Learning for Computer Vision in Hall in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Lunch DTSTART:20160731T074500Z DTEND:20160731T084500Z DTSTAMP:20260421T155209Z UID:session/STnX1Rq5ebKCDaG6ya9JRF@hasgeek.com SEQUENCE:0 CREATED:20160623T160046Z DESCRIPTION:\n LAST-MODIFIED:20160706T085826Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Lunch in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Building a scalable Data Science Platform ( Luigi\, Apache Spark\, Pandas\, Flask) DTSTART:20160731T084500Z DTEND:20160731T103000Z DTSTAMP:20260421T155209Z UID:session/zjxAZwsDfYJyGpxb2Yn7i@hasgeek.com SEQUENCE:2 CATEGORIES:Workshop,Intermediate CREATED:20160623T155602Z DESCRIPTION:The biggest challenge in building a data science platform is t o glue all the moving pieces together. Typically\, a data science platform consists of:\n\n* Data engineering - involves a lot of ETL and feature en gineering.\n* Machine learning - involves writing a bunch of machine learn ing models and persistence of the model\n* API - involves exposing end p oints to the outside world to invoke the predictive capabilities of the mo del\n\nOver time the amount of data stored that needs to be processed incr eases and it necessitates the need to run the Data Science process frequen tly. But different technologies/stack solve different parts of the Data Sc ience problem. Leaving it to respective teams introduces lag into the syst em. What is needed is an automated pipeline process - one that can be invo ked based on business logic (real time\, near-real-time etc) and a guarant ee that it will maintain data integrity.\nDetails of the workshop\n\n### D ata Engineering\n\nWe all know that 80% of the effort is spent on data eng ineering while the rest is spent in building the actual machine learning m odels. Data engineering starts with identifying the right data sources. Da ta sources can be databases\, third party APIs\, HTML documents which need s to be scrapped and so on. Acquiring data from databases is a straight fo rward job\, while acquiring data from third party APIs and scrapping may c ome with its own complexities like page visit limits\, API rate limiting e tc. Once we manage to acquire data from all these sources\, the next job i s to clean the data. \n\nWe will be covering the following topics for data engineering:\n\n* Identifying and working with 2 data sources.\n* Writing ETL (Extraction\, Transformation and Loading) with Pandas\n* Building de pendency management with Luigi\n* Logging the process\n* Adding notificati ons on success and failure\n\n### Machine Learning\nBuilding a robust and scalable machine learning platform is a hard job. As the data size increas es\, the need for more computational capabilities increase. So how do you build a system that can scale by just adding more hardware and not worryin g about changing the code too much every time? The answer to that is to us e Apache Spark ML. Apache Spark lets us build machine learning platforms b y providing distributed computing capabilities out of the box.\n\nWe will be covering the following topics for Machine Learning:\n\n* Feature Engine ering\n* Hypothesis to solve\n* Configuration of environment variables for Apache Pyspark\n* Build the Machine Learning code with Apache Spark\n* Pe rsisting the model\n\n### API\nIt ain't over until the fat lady sings. Mak ing a system API driven is very essential as it ensures the usage of the b uilt machine learning model \, thereby helping other systems integrate the capabilities with ease. \n\nWe will be covering the following topics for API:\n\n* Building REST API with Flask\n* Based on the input parameters\, build respective methods to extract features to be fed into the model\n* S end responses as a JSON\n\nPre-Requisites:\n\n* Knowledge of python\n* Bas ic understanding of Data science\n\n### Speaker bio\n\nSpeaker Bio:\n\n**N ischal** is co founder and Data Engineer at Unnati Data Labs who enables t he Data Scientists to work at peace. He makes sure that they get the data they need and in the way they need it. Previously he has built\, from scra tch\, various systems for E-commerce like catalog management\, recommendat ion engines and market basket analysis to name a few during his tenure at Redmart.\n\n**Raghotham** is a co founder Data Scientist at Unnati Data La bs\, who can work across the complete stack. Previously\, at Touchpoints I nc.\, He single handedly built a data analytics platform for a fitness wea rable company. With Redmart\, he worked on the CRM system and has built a sentinment analyzer for Redmart’s Social Media. Prior to Redmart and Tou chpoints\, Raghotham worked at SAP Labs where he was a core part of what i s currently SAP’s framework for building web and mobile products. He was a part of multiple SAP wide events helping to spread the knowledge both i nternally and to customers.\n\nThey have conducted workshops in the field of Deep learning across the world. They are strong believers of open sourc e and love to architect big\, fast and reliable systems.\n LAST-MODIFIED:20230810T072606Z LOCATION:Hall - TERI\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/2016/schedule/building-a-scalable-da ta-science-platform-luigi-apache-spark-pandas-flask-zjxAZwsDfYJyGpxb2Yn7i BEGIN:VALARM ACTION:display DESCRIPTION:Building a scalable Data Science Platform ( Luigi\, Apache Spa rk\, Pandas\, Flask) in Hall in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Tea Break DTSTART:20160731T103000Z DTEND:20160731T104500Z DTSTAMP:20260421T155209Z UID:session/KzzhEjaL8ukrFV2hxyC78j@hasgeek.com SEQUENCE:0 CREATED:20160623T160203Z DESCRIPTION:\n LAST-MODIFIED:20160706T085928Z LOCATION:Bangalore ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Tea Break in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT BEGIN:VEVENT SUMMARY:Contd. Building a scalable Data Science Platform ( Luigi\, Apache Spark\, Pandas\, Flask) DTSTART:20160731T104500Z DTEND:20160731T123000Z DTSTAMP:20260421T155209Z UID:session/1221giHLznx3Pd5X8W5mps@hasgeek.com SEQUENCE:0 CREATED:20160623T155656Z DESCRIPTION:\n LAST-MODIFIED:20160726T055258Z LOCATION:Hall - TERI\nIN ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com BEGIN:VALARM ACTION:display DESCRIPTION:Contd. Building a scalable Data Science Platform ( Luigi\, Apa che Spark\, Pandas\, Flask) in Hall in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT END:VCALENDAR