Neural Networks For Feature Analysis

Introductions

Mayank Barad
Rising Senior in Computer Engineering and Computer Science

Daksh Khetarpaul
Rising Junior in Computer Engineering

Katherine Lew
Rising Sophomore in Finance and Computer Science

Advisors - Dr Richard Howard, Dr Richard Martin

Project Description

Neural networks have a long history of being used for classification, and more recently. content generation, Example classifiers including, image classification between dogs and cats, text sentiment classification. Example generative networks include those for human faces, images, and text. Rather than classification or generation, this work explores using networks for feature analysis. Intuitively, features are the high level patterns that distinguish data, such as text and images, into different classes. Our goal is to explore bee motion datasets to qualitatively measure the ease or difficulty of reverse-engineering the features found by the neural networks.

Week 1

• Understanding the purpose of the project

• Setting up Github and iLab accounts

• Getting familiar with Neural Networks

Week 2

• Visited the beehive to observe the behavior of real bees

• Made a prototype simulator with pygame - Rejected(pretty obvious reasons)

• Integrated "Power Law" for a more natural bee motion

First Prototype → Applying "Power Law" →

Week 3

• Programmed a randomness function that allows the user to adjust the degree of randomness of synthetic bee motion along a spectrum, where 0.0 represents a "bee" moving in a completely random motion and 1.0 represents a "bee" moving via a distinct non-random pattern (e.g in a clockwise circle)

• Trained machine learning model (AlexNet) to identify random versus non-random behavioral patterns using the randomness function

• Researched Shannon's Entropy Function as measure of model accuracy and created program that calculates the joint entropy of two discrete random variables within the randomn system (e.g angle and distance)

Week 4

Week 5