AI Research: Bayesian ML, Causal ML, RL, Graph ML

Kurs haqida

Kurs zamonaviy AI sohasida faol rivojlanayotgan to'rtta tadqiqot yo'nalishi bilan tanishtiradi: Bayesian ML, Causal ML, Reinforcement Learning va Graph ML. Har bir yo'nalish nazariy asoslari, dolzarb ochiq masalalari va real muammolar bilan bog'liqligi nuqtai nazaridan ko'rib chiqiladi. Kurs magistrlik dissertatsiyasi mavzusini tanlash uchun poydevor bo'lib xizmat qiladi.

Nimalarni o'rganasiz

✓ELBO ni chiqarish va generatsiya masalasi uchun variatsion avtoenkoder (VAE) ni amalga oshirish

✓Sabab-oqibat grafini (causal graph) qurish va interventsion effektlarni baholash uchun do-calculus qo'llash

✓Nazorat masalasida RL-agentni PPO usuli bilan o'qitish va yaqinlashishni (convergence) tahlil qilish

✓Message passing asosida GNN ni amalga oshirish va uning ifoda quvvatini (expressive power) baholash

Asosiy mavzular

•Bayesian ML: variatsion xulosa chiqarish (VI), MCMC, chuqur bayes modellari (BNN, VAE)

•Causal ML: sabab-oqibat graflari (DAG), do-calculus, counterfactual reasoning

•Kuzatuv ma'lumotlaridan causal inference: propensity score, instrumental variables

•Reinforcement Learning: MDP, policy gradient (REINFORCE, PPO), Q-learning (DQN)

•LLM da RL: RLHF, GRPO, reward hacking

•Graph Neural Networks: GCN, GAT, message passing, expressive power (WL test)

•Geometrik Deep Learning: invariantlik va ekvivariantlik

•Yo'nalishlar orasidagi bog'liqlik: bayes RL, causal representation learning

Ushbu tavsif avtomatik tarzda yaratilgan va hali o'qituvchi tomonidan tekshirilmagan — bu muhokama uchun qoralama.

Kurs haqida

Nimalarni o'rganasiz

Asosiy mavzular

Fikr-mulohaza