xcorr-notebooks

An ongoing collection of ipython notebooks and interactive visualizations on neuroscience, machine learning & computer science from xcorr: computational neuroscience.

By Patrick Mineault, PhD

Machine learning

• ICA with normalizing flow models (NICE)
• PCA through gradient descent
• Gibbs sampling & block Gibbs sampling for the Ising model
• Persistent contrastive divergence for maximum entropy models

Reinforcement learning

• Contextual bandits with Thompson sampling
• Q-Learning in OpenAI gym
• Multi-armed bandit as a Markov decision process

Systems identification and Wiener-Volterra analysis

• Second order Wiener-Volterra estimation
• Reduced variance in reverse correlation estimates, antithetic sampling
• Polynomial regression vs. Taylor expansion

Interactive visualizations and widgets (Javascript)

• Linear integrate-and-fire neurons
• Networks of balanced E/I linear integrate-and-fire neurons
• A scrollytelling tour of balanced E/I networks
• Optimal control with a linear quadratic regulator
• Stabilizing the cart pole with control theory
• Self-organized criticality: the forest fire model
• RGB color mixing
• Nonlinear feedback to decouple network of oscillators

Complex systems

• Network analysis of C. Elegans (Julia)
• Percolation model
• Self-organized criticality in an Abelian sandpile model (Julia)

Data Science

• What happened to my body when I broke a leg (Garmin watch data)

Utils

• Render OpenAI gym as GIF