Collection of possibly not self-contained notes written up for myself on various topics that I have investigated.

A Better Proof of Unbiasedness of the Sequential Monte Carlo Based Normalizing Constant Estimator (2023/06/22)
AI existential risk (2023/06/05)
Bird’s eye view of Large Language Models (2023/05/15)
Human Compatible (2023/04/24)
Two Views of Sequential Monte Carlo (2022/11/18)
Reversible jump Markov chain Monte Carlo (2022/11/11)
Optimal Smoothing Sequential Monte Carlo Proposal for Linear Gaussian State Space Model (2022/08/08)
Filtering and smoothing in HMMs and LGSSMs (2022/07/16)
Build a baby and let it grow (2021/12/04)
Independent Component Analysis (2021/08/25)
Learning and amortized inference in hierarchical Bayesian models (2020/09/04)
Neurosymbolic generative models (2020/07/17)
Alternative proof for tighter bounds (2019/12/17)
Semi-supervised model learning and amortized inference (2019/09/27)
Conditional Variational Autoencoders (2018/06/27)
Convergence of Gradient Descent (2018/03/14)
Variational Inference for Monte Carlo Objectives (VIMCO) (2018/02/25)
Neural Variational Inference and Learning in Belief Networks (NVIL) (2018/02/25)
Reweighted Wake-Sleep (2018/02/21)
Attend, Infer, Repeat (2018/02/19)
Neural Adaptive Sequential Monte Carlo (2018/02/19)
\(n\)-step Relationships for \(Q\) Function and Value Function (2018/01/16)
Deep Q Network (DQN), Policy Gradients, Advantage Actor Critic (A2C) (2018/01/13)
REBAR and RELAX (2017/12/26)
Amortized Inference (2017/12/19)
Reverse vs Forward KL (2017/12/17)
Hierarchical Reinforcement Learning (2017/10/22)
Bounded Optimality (2017/10/22)
Optimal Sequential Monte Carlo Proposal for Linear Gaussian State Space Model (2017/09/21)
Heuristic Factors (2017/09/15)
Resampling Algorithms (2017/09/15)
Effective Sample Size (2017/09/14)
Importance Sampling (2017/09/14)
Curse of Dimensionality (2017/09/11)
reparameterization trick for joint random variables (2017/08/10)
gaussian processes for regression (2017/06/20)
analysis-by-synthesis by learning to invert generative black boxes (2017/04/19)
variational lossy autoencoder (2017/04/19)
Unbiasedness of the Sequential Monte Carlo Based Normalizing Constant Estimator (2017/04/05)
information theory basics (2017/04/04)
structured inference networks for nonlinear state space models (2017/03/13)
why is sequential monte carlo better than importance sampling? (2017/03/13)
probabilistic view of ai (2017/03/10)
utility theory (2017/03/10)
automatic statistician (2017/02/27)
c19 unsupervised machine learning lecture notes (2017/02/21)
Occam’s Razor and Model Selection (2017/02/21)
importance weighted autoencoders (2017/02/19)
box’s loop (2017/02/14)
improved semantic representations from tree-structured long short-term memory networks (2017/02/13)
making neural programming architectures generalize via recursion (2017/02/10)
Variance Reduction Techniques (2017/02/04)
black box variational inference (2017/02/04)
stochastic variational inference (2017/01/16)
variational inference with normalizing flows (2017/01/10)
c3: lightweight incrementalized mcmc for probabilistic programs using continuations and callsite caching (2017/01/02)
Planning by dynamic programming (2016/11/08)
Concrete distribution (2016/11/07)
REINFORCE trick (2016/11/07)
Measure theory for machine learning (2016/09/29)
The Expectation-Maximization algorithm (2016/09/28)
Bellman equations for reinforcement learning (2016/09/24)
Reparameterization trick (2016/09/05)
Variational autoencoders (2016/09/05)
Gaussian unknown mean (2016/09/04)
old bayesian machine learning notes (2014/08-2015/06)