Conference: March 5-7, 2018

Workshops: March 8-9, 2018

Presentation: Neural Networks Across Space and Time

Track: Evolving Java and the JVM: Mobile, Micro and Modular

Location: Mountbatten, 6th flr.

Duration: 10:35am - 11:25am

Day of week: Wednesday

Level: Intermediate

More talks on:

Share this on:

What You’ll Learn

  • Hear what neural networks are and what types of deep learning networks are out there.
  • Learn how to decide if your solution needs a convolutional network or a recurrent one.
  • Find out what some of the DL resources are out there and see how some are used.

Abstract

This is an intro-level talk on deep learning. We’ll start with a brief introduction to deep neural networks, why they are important and how they work. The talk will then cover two of the most important deep neural architectures: convolutional networks which excel at handling images and recurrent networks which handle time-series or sequential input (such as text). We'll show examples of both convolutional and recurrent networks using the deeplearning4j framework.

Question: 

What is the focus of your work today?

Answer: 

I’m one of the lead developers of the management layer for Horizon Cloud; VMware’s cloud-hosted virtual desktop product

Question: 

What’s the motivation for this talk?

Answer: 

Andre Karpathy has described neural networks as "Software 2.0" Even if he's overestimating, machine learning is only going to get more important. Developers wanting to stay ahead should at least gain a basic understanding of what's involved and how it can be used.

Question: 

How you you describe the persona and level of the target audience?

Answer: 

A developer who's heard some of the hype about deep learning and wants to know what deep learning actually is. No prior knowledge of machine learning is assumed.

Question: 

What do you want this persona to walk away from your talk knowing that they might not have known 50 minutes before?

Answer: 

A basic understanding of what a neural network is and how it works. The two predominant deep network architectures today, what they are used for and how to build them.

Question: 

This talk is in the Java track. Do you think this talk would be interesting to non-java developers as well?

Answer: 

Yes, because although I'm going to use Java examples, a lot of frameworks out there have broadly similar primitives.

Speaker: Dave Snowdon

Staff Engineer @VMware

By day, Dave Snowdon is a mild-mannered programmer working on cloud management of virtual desktop infrastructure (VDI) at VMware. By night, when not asleep, he plans world domination by social emotional robots powered by python and clojure. Before he was virtualised Dave worked for Xerox Research in France and, back in the mists of time, developed one of the first distributed multi-user virtual reality environments as part of his PhD work at Manchester. Dave wrote the software for Pete Townshend's (of The Who) Lifehouse Method - a system to generate unique pieces of music for each user based on Information extracted from audio, rhythms and images supplied by the user. More recently, he has been devoting his time to understanding machine learning with a particular emphasis on deep neural networks.

Find Dave Snowdon at

Speaker page

@davesnowdon
Dave Snowdon

Last Year's Tracks

Monday, 5 March

Tuesday, 6 March

Wednesday, 7 March

This Year's Schedule

Learn trends from innovator and early adopter companies that you can bring home to your team

Sign up to be informed when registration for QCon London 2019 (Mar 4-8) opens!

Sign Up