Exploring Machine Learning Research Papers

Throughout the summer of 2024 I am exploring various machine learning papers, and implementing the tools that they develop. Doing so I am aiming to develop and intuition of the properties of machine learning models. Specifically, I am interested in implementing tools that elucidate the geometry of neural networks and give insights to what is going on in a neural network. The accompyning code for these posts can be found here.

• SplineCam: Exact Visualization and Characterization of Deep Network Geometry and Decision Boundaries
• Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV)
• Linearity of Relation Decoding in Transformer Language Models
• Characterizing the Decision Boundary of Deep Neural Networks