arrow_back Making Deep Neural Networks smaller and faster
Applied Deep Learning
Submitted by Abhishek Thakur (@abhishekthakur) on Sunday, 29 May 2016
Section: Full talk Technical level: Intermediate
This talk gives an overview of current advancements in the field of deep learning and neural networks and explains how neural networks can be implemented and used for most of the machine learning problems.
The talk consists of an introduction to deep learning followed by live examples on the implementation of deepnets using python. The talk then discusses how to optimize the parameters of a neural network and how to carefully choose the hyperparameters. Afterwards, we move to pretrained networks and methods of finetuning a pretrained network. The talk will show use of python with keras and Caffe.
I currently work as a Senior Data Scientist at Searchmetrics Inc. in Berlin, Germany.
I have participated in over 100 machine learning competitions and the knowledge acquired by them are put to use in my daily job. I also developed an AutoML system based on the knowledge gained from the wide range of datasets Ive worked with. The AutoML system recently won the final phase of AutoML cpu track and the neural network framework won the only GPU track of the competition.
More information about me can be found at: www.bit.ly/thakurabhishek and on www.kaggle.com/abhishek