1 / 24

Introduction

This project aims to develop an incrementally learning artificial neural network (ANN) that can adapt to dynamic systems without catastrophic forgetting. The end product includes a GUI for interacting with the ANN and modeling nonlinear systems, such as power load forecasting and market trend prediction. The project focuses on creating a software design document, implementing incremental learning capabilities, and providing comparative analysis with traditional ANNs. The technical approach involves using augmenting neural nets and exploring different design alternatives to enhance learning efficiency. The evaluation of project success highlights meeting first-semester milestones and outlines recommendations for future work. Lessons learned emphasize effective team communication and time management strategies. This initiative opens up new possibilities in the field of artificial intelligence and neural network development.

tpaulus
Download Presentation

Introduction

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Introduction Incremental Learning With Neural Networks February 1, 2001 May01-14 Team Members David Herrick Brian Kerhin Chris Kirk Ayush Sharma Clients/Faculty Advisors Dr. Eric Bartlett

  2. Overview Overview • Problem Statement (Ayush) • Design Objectives (Ayush) • Intended Users (Ayush) • End-Product Description (Brian) • Assumptions/Limitations (Brian) • Project Risks & Concerns (Brian) • Technical Approach (Brian) • System Overview (Dave) • Technical Design (Dave) • Evaluation of Project Success (Chris) • Possible Future Work (Chris) • Human/Financial Budget (Chris) • Lessons Learned (Chris) • Closing Summary (Chris)

  3. ANN Description ANN Description = • Real neurons are decision making cells in the brain • Software neurons function similarly, to make decisions in software • The brain is a network of “Real neurons” making decisions in parallel • ANNs accomplish the same goal using “Software neurons” • ANNs are used to interpolate nonlinear systems that are very complex • Predict trends in the Stock Market • Predict trends in power use through out the year • Predict trends in global whether patterns

  4. Problem Statement Problem Statement Unlike humans, who incrementally learn information as it is introduced, ANNs learn all at once. An existing ANN cannot adapt to dynamically changing system (i.e. CATASTROPHIC FORGETTING). Hence, as a system changes, a new ANN must be created from scratch. For many real-life applications of ANNs, it is impractical to regularly create replacement ANNs.

  5. Design Objectives Design Objectives • Create a software design document • Create an incrementally learning ANN • Create a GUI for incrementally learning ANN • Apply to a power load problem and compare to traditional ANNs

  6. Operation and Construction Environment Hardware and Software (Supplies) • Hardware • Intel Gateway PCs in Adaptive Computing Laboratory • Software • Microsoft Windows NT4.0 • Microsoft Visual C++ • Microsoft Visual Basic

  7. Intended Users Intended Users • Adaptive Computing Laboratory • Dr Eric Bartlett • Research Assistants • Other Neural Network Programmers

  8. End-Product Description End-Product Description • Incrementally learning ANN to model dynamic nonlinear systems • Learning System can learn new data without forgetting • Has goodness measures • GUI operated, via command line

  9. Assumptions Assumptions • User should have a basic knowledge of neural networks • Data file will be tab delimited following Dr Bartlett’s specified file format

  10. Limitations Limitations • Set time period to produce code and documents • Number of developers is not based on commercial properties of the software • Computationally intensive technique of problem solving

  11. Project Risks and Concerns Project Risks & Concerns • Major accidents can (and have) cripple team members • Finishing on time • Incremental learning may not improve goodness measures or error results

  12. Technical Approach Technical Approach (Design Alternatives) • One strong neural network with augmenting neural nets • Several augmenting neural nets • Update a single augmenting neural net • Produce result based on outputs average • Produce result based on output sums • Multiple weak neural nets average results

  13. System Overview Incremental Learning System

  14. Technical Design Technical Design

  15. Technical Design (cont.) Technical Design

  16. System Overview Incremental Learning System

  17. Evaluation of Project Success Evaluation of Project Success • 1st Semester Milestones • Project Plan (Fully Met) • Project Poster (Fully Met) • Design Report (Fully Met) • Hardware Requirements (Fully Met) • Software Requirements (Fully Met) • Software Design (Fully Met)

  18. Evaluation of Project Success (cont.) Evaluation of Project Success (cont.) • 2nd Semester Milestones • Software Implementation (Partially Met) • 100% of algorithm designed • 10% of algorithm implemented • Final Implementation (Not Met) • Final Report (Not Met) • Presentation for Industrial Review Panel (Partially Met) • 80% of Presentation Completed

  19. Possible Future Work Recommendations for Further Work • Explore alternative learning styles • More research tools in the field of artificial intelligence

  20. Human Budget Human Budget

  21. Financial Budget Financial Budget

  22. Lessons Learned Lessons Learned • Establish two weekly meetings • Progress meeting with faculty advisors • Development meeting with team members • Wounded team members don’t improve group efficiency • Keep faculty advisors well-informed of progress and seek feedback • Plan to complete milestones ahead of schedule • Balance workload among team members

  23. Closing Summary Closing Summary • New way of looking at neural networks • Overcomes limitations of traditional neural nets • Can be greatly reused and built upon • Will further the field of Artificial Intelligence

  24. Questions? Questions?

More Related