machine-learning-theory

A classified list of meta learning papers based on realm.

[ICML2026] The first, fully verified, sorry-free, large-scale Lean 4 library for statistical learning theory, covering infrastructures for mordern statistics and learning theory.

Learn the theory, math and code behind different machine learning algorithms and techniques.

This article reframes pricing as a negotiation rather than a prediction, showing how price emerges from tensions between product reality, market dynamics, and buyer behavior. It introduces negotiation-aware ML, value decomposition, and equilibrium modeling to build transparent, human-aligned pricing systems.

This article explores the theory behind explainable car pricing using value decomposition, showing how machine learning models can break a predicted price into intuitive components such as brand premium, age depreciation, mileage influence, condition effects, and transmission or fuel-type adjustments.

[NeurIPS 2023] The official implementation of "Rethinking Semi-Supervised Imbalanced Node Classification from Bias-Variance Decomposition" .

A 4-skill pipeline for Claude Code: verify mathematical proofs → repair with literature support → sharpen the theory → write corrected proofs. Integrates Codex MCP for adversarial cross-review. Venue-audited reference library across statistics/econometrics/ML theory.

This repository hosts a progressive series of implementations (Code_v1, Code_v2, and beyond) for deterministic β*-optimization in the Information Bottleneck framework. Includes symbolic fusion, multi-path inference, and Alpay Algebra–driven critical point validation (β* = 4.14144).

Implementation is to use gradient descent to find the optimal values of θ that minimize the cost function.

ML course project: investigation on common perceptions of same neural network model with different random seed

A growing list of papers, books, courses and blogs related to machine learning theory and optimization

📓 Chapter summaries adapted from the textbook "Understanding Machine Learning"

Official code repository for "Distributional Autoencoders Know the Score", NeurIPS 2025

A clean, theoretical implementation of Stochastic Variance Reduced Gradient (SVRG) demonstrating linear convergence rates on convex objectives. Includes comparative analysis against vanilla SGD with empirical validation on synthetic and real-world datasets.

Emergent Computational Epistemology: studying AI’s emergent behaviors as non-human epistemic systems.

Towards understanding Machine Learning theory in 69 days :)

My personal projects of re-learning machine learning and deep learning algorithms

Code and reproducibility artifact for projective root anti-concentration and graph-learning applications.

Gradient resurrection: a theoretically-derived, trust-scaled controller for continual learning, with full proof and an honestly-reported failed extension hypothesis