Revisit: Analyzing the Stock Market with Transfer Entropy Graphs
Eric Taylor Eric Taylor

Revisit: Analyzing the Stock Market with Transfer Entropy Graphs

This project revisits using transfer entropy graphs to predict stock market volatility, replacing a complex prior method with a simpler feature-based approach. By analyzing many historical graphs and extracting structural and volatility features, it seeks to add predictive power beyond recent price trends. Early results show improved performance, especially when classifying volatility levels instead of predicting exact values.

Read More
Using Static Code Metrics to Model LLM Test Creation Ability
Eric Taylor Eric Taylor

Using Static Code Metrics to Model LLM Test Creation Ability

As companies increasingly rely on Large Language Models (LLMs) for code-related tasks, optimizing model selection becomes crucial to balancing performance and cost. This study explores whether static code metrics can predict an LLM's ability to generate effective test cases. By analyzing various Java data structures, this project correlates static complexity metrics with test generation success and code coverage.

Read More
Using Reinforced Learning to Create Efficient Random Tests
Eric Taylor Eric Taylor

Using Reinforced Learning to Create Efficient Random Tests

A way to use Reinforced Learning to generate more efficient stimulus to not only get more coverage per test, but also cover more sooner within each test. At the same time keep some randomness to have the ability to generate an efficient "Smart" Random test suite that can hopefully replace some bulky pure random tests.

Read More
Using Q-Reinforced Learning to Generate Digital Logic
Eric Taylor Eric Taylor

Using Q-Reinforced Learning to Generate Digital Logic

This article goes into my experiment applying Reinforcement Learning to craft digital logic. Discover the surprising effectiveness of translating Q function tables into logic, offering real-world implementation insights.

Read More