SPA Conference session: Advanced Problem Solving: Neural Nets
|One-line description:||Neural nets are a sophisticated means of training a computer how to solve though problems.|
|Session format:||Tutorial 150 minutes [read about the different session types]|
|Abstract:||With the introduction of [design patterns][patterns] developers gained|
a powerful tool. Patterns aid in the communication about
solutions to a problem and expand the repertoire of design-problem solving
Design patterns concentrate on software design. They offer no advice
on solving a problem in the first place. By studying advanced problem
solving techniques, a developer gains a similar powerful tool.
In this session we take a closer look into neural nets. Neural nets
are able to figure out how to solve a specific task. By training a
Neural net it figures what the correct responses are given some stimuli.
After an introduction to neural nets, you will get the opportunity to
try your hands on a few problems of your own. We even go the extra
mile and implement neural nets our selves.
|Audience background:||Interest in solving problems.|
|Benefits of participating:||By participating you will benefits in the following ways:|
* Acquire a powerful tool in you problem solving toolkit.
* Be able to create, verify and train neural nets.
* Acquire a practical set references for future investigations.
|Materials provided:||* Presentation|
* Environment to experiment with neural nets.
* Interesting problems to solve.
* A guiding set of failing test to implement a neural nets.
|Process:||The session will start of with a practical introduction to neural nets|
During the introduction examples of neural nets and how the are able
to solve complex problems are given.
After the introduction the participants can try their hands on
training neural nets them selves.
After getting to know the neural net paradigm, participants can test
drive a custom implementation of neural nets. Participants can work of a provided scaffolding of code, or use the specifications to drive their own implementation.
There will be two types of exercises.
1. Specific problems that with neural nets.
2. A guided TDD session to implement neural nets.
It is advisable to team up and pair-programming throughout the
practical part of the session.
|Detailed timetable:||00:00 - 00:45 Presentation about neural nets and how to use them.|
00:45 - 01:00 Introduction to several problems
01:00 - 01:30 Solving problems with neural nets
01:30 - 02:30 TDD a custom implementation for a neural net framework
|Outputs:||Working knowledge of neural nets and the means to implement a framework. Handy set of reference for production-ready frameworks.|
|History:||This will be the first time I will give this session.|
|1. Daan van Berkel