Deep Learning, neon, and the Nervana Cloud

*Note, expedite Check in at Galvanize; register here (https://www.eventbrite.com/e/sf-data-science-meetup-deep-learning-neon-and-the-nervana-cloud-tickets-21567756741)

Building and deploying deep learning models holds tremendous potential to revolutionize every industry. Using hands-on exercises and tutorials, this meetup will showcase how deep learning can be quickly applied to image, video and natural language processing problems with the aim of providing participants the knowledge they would need to get started on their datasets.

We have one main talk and three lightning talks lined up for the evening :)

Agenda:
6:00 - Doors open - Networking & Food.
6:30 - Intro to Deep Learning with Urs Köster, Machine Learning Engineer @ Nervana
7:00 - Interactive neon workshop with Nervana Machine Learning Experts
7:30 - Lightning talk (10 min): "Bayesian Optimization of Deep Learning Models using SigOpt with neon & ncloud” by Scott Clark, CEO @ SigOpt
7:45 - Lightning talk (10 min): “Object Detection and Localization” by Yinyin Liu, Machine Learning Engineer @ Nervana
8:00 - Lightning talk (10 min): “Activity Recognition in Videos” by Sathish Nagappan, Machine Learning Engineer @ Nervana
8:15 - Q&A & wrap-up.
8:30 - Networking
9:00 - Doors close

What to Bring:
Remember to bring your laptop so you can participate in the neon workshop! If you want to install neon to your laptop, you can download it from GitHub at https://github.com/nervanasystems/neon. Cloud instances will also be provided.

Detailed Agenda & Speaker Profiles:

Main Talk: Deep Residual Nets, Activity recognition in videos, and Q&A systems using neon and the Nervana Cloud

Talk #1: Deep Residual Nets, Activity recognition in videos, and Q&A systems using neon and the Nervana Cloud

Will Constable will start with an introduction to the field of Deep Learning, neon and the Nervana Cloud. The presentation will be followed by an interactive workshop using neon. neon is an open-source Python based Deep Learning framework that has been built from the ground up for speed, scalability and ease of use.

Speaker: Will Constable, Machine Learning Engineer at Nervana

Will has over four years of experience at Qualcomm developing software for autonomous robots, neural networks, computer vision applications and event-based neuromorphic sensors. He enjoys working with interdisciplinary teams solving complex problems in software and hardware.

Lightning Talk #1: Bayesian Optimization of Deep Learning Models using SigOpt with neon & ncloud

In this talk, Scott Clark will show how the Bayesian Optimization methods used by SigOpt, coupled with the incredibly scalable deep learning architecture provided with the Nervana Cloud and neon, allow anyone to easily tune their models to quickly achieve higher accuracy. Scott will walk through the techniques and show an explicit example with results.

Speaker: Scott Clark (https://www.linkedin.com/in/sc932), CEO at SIgOpt

Scott has been applying optimal learning techniques in industry and academia for years, from bioinformatics to production advertising systems. Before SigOpt, Scott worked on the Ad Targeting team at Yelp leading the charge on academic research and outreach with projects like the Yelp Dataset Challenge and open sourcing MOE. Scott holds a PhD in Applied Mathematics and an MS in Computer Science from Cornell University and BS degrees in Mathematics, Physics, and Computational Physics from Oregon State University. Scott was chosen as one of Forbes' 30 under 30 in 2016.

Lightning Talk #2: Object Detection and Recognition

The second lightning talk will be presented by Yinyin Liu. She will present a model for object detection and localization, called Fast-RCNN. She will show how to introduce a ROI pooling layer into neon, and how to add the PASCAL VOC dataset to interface with model training and inference. Lastly, Yinyin will run through a demo on how to apply the trained model to detect new objects.

Speaker: Yinyin Liu (https://www.linkedin.com/in/yinyin-liu-0228986), Machine Learning Engineer at Nervana

Yinyin has over ten years of experience in developing neural models and learning algorithms, applying computational neuroscience, and machine learning to solve problems in areas including image recognition, reinforcement learning and robotics. She recently worked on NLP problems using recurrent neural networks at Nervana.

Lightning Talk #3: Video Activity Recognition and NLP Q&A Model Example

In this talk, Sathish Nagappan will present an introduction to the UCF-101 video activity recognition dataset and discuss how 3-D convolutions work. A demo will be presented on how to predict actions in video clips. Lastly, an NLP Q&A model example will be presented.

Speaker: Sathish Nagappan (https://www.linkedin.com/in/sathish-nagappan-794a1a45), Machine Learning Engineer at Nervana

Sathish is a recent Stanford graduate where he did research in Andrew Ng’s machine learning group. He is very entrepreneurial, has great design sense, and has experience across the entire engineering stack from hardware to software. He is really interested in art, design, AI, travel, movies, and Baltimore Ravens football.