How to sample from a distribution
How to perform the seemingly simple task of sampling from a distribution.
Posts by Year
2024
Notes for Generative AI learning, part 2
This post is for chapter 5 (Autoregressive Models) and chapter 6 (Normalizing Flow Models).
Notes for Generative AI learning, part 1
Scratch notes when reading the O’Reilly book of ‘Generative Deep Learning, 2nd Edition’. This post is for the first 4 chapters.
2023
Large Language Model (LLM) learning notes, part 2
Learning notes from the Coursera course on Large Language Models (LLMs)
Large Language Model (LLM) learning notes, part 1
Learning notes from the Coursera course on Large Language Models (LLMs)
Notes on Reinforcement Learning: Planning
A collections of notes of Reinforcement Learning, as I am going through the Coursera specialization: Fundamentals of Reinforcement Learning. Hopefully this w...
Notes on Reinforcement Learning: Temporal Difference for control
A collections of notes of Reinforcement Learning, as I am going through the Coursera specialization: Fundamentals of Reinforcement Learning. Hopefully this w...
How to use US Census Bureau data
A quick primer on how to leverage the rich data from the US Census Bureau.
Notes on Reinforcement Learning: Temporal Difference learning for prediction
A collections of notes of Reinforcement Learning, as I am going through the Coursera specialization: Fundamentals of Reinforcement Learning. Hopefully this w...
2022
Notes on Reinforcement Learning: Monte Carlo method
A collections of notes of Reinforcement Learning, as I am going through the Coursera specialization: Fundamentals of Reinforcement Learning. Hopefully this w...
Notes on Reinforcement Learning: Policy and value iterations
A collections of notes of Reinforcement Learning, as I am going through the Coursera specialization: Fundamentals of Reinforcement Learning. Hopefully this w...
Notes on Reinforcement Learning: MDP
A collections of notes of Reinforcement Learning, as I am going through the Coursera specialization: Fundamentals of Reinforcement Learning. Hopefully this w...
Notes on Reinforcement Learning: Bandit
A collections of notes of Reinforcement Learning, as I am going through the Coursera specialization: Fundamentals of Reinforcement Learning. Hopefully this w...
Notes on optimization: the basics
A series of study notes of optimization, including convex optimization, linear programming, linear simplex method, and KKT conditions.
2021
MLOps course: Introduction
Summary of Coursera MLOps Course 1.
Book roundup, February 2021
Two diagonally different topics, but equally enlightening books.
2020
Machine Learning Design Patterns: Model Training
Here we look into a good resource of practicing good machine learning design patterns.
Machine Learning Design Patterns: Reproducibility
Here we look into a good resource of practicing good machine learning design patterns.
Statistical tests for multiple comparisons
We are often spoiled by the amount of data at our disposal, but it is still easy to make mistakes when making statistical inferences if we exploit too much.
The model to flatten the curve
With the covid-19 pandemic distruping everyday life across the globle, a phrase we hear a lot is to “flatten the curve”. What is the science behind that?
Multi-armed bandit
Multi armed bandit is such a classic problem, here let’s implement some simple policies from ground up to address this problem.
Book roundup, January 2020
Statistics, how to build intuition to see everything from a different angle.
2019
The egg drop problem
Egg drop soups are delicious, dropping eggs can also be fun.
The secretary problem
How to maximize your chance to find the best candidate, apartment, or even soulmate, only if the world can be modeled simply.
Book roundup, July 2019
Self-improvement, in difference aspects.
Containers and their orchestration
Recap of an excellent 3-day workshop on Docker and Kuberentes.
Model as a Service through Flask
How to speed up the time from between model development to model deployment.
Book roundup, February 2019
Math, debt, and Amazon.
Welford algorithm for updating variance
How can one update the mean, variance, and median of a long list of numbers, after a new element is added?
2018
Spectral clustering, step by step
How can one relate a seemingly obivous physics concept to linear algebra, and then to a data science problem.
Pivot a Hive table
There is no built-in pivot function in Hive, but one can still do it with relative ease.
Schema on write v.s. schema on read
A runner’s view on this database concept.
Platt scaling for probability calibration
For classification problems, sometimes we care about the narrative of the predicted scores more than the predicted class. But the predicted scores can not al...
Gradient descent with the right step
Gradient descent is one of the most important tools in machine learning, but how hard can it be?
Book roundup, August 2018
How to lie, to get rich, with statistics, and to play people’s mind.
LIME, anyone?
Why Should I Trust Your black-box model? How about breaking open the black-box model, at least locally.
Book roundup, June 2018
Armed with seemingly omnipotent technical marvels that are destined to solve all the problems should they arise, there are still questions we need to think c...
Fitting a Mixture of Gaussians
Fitting one single Gaussian distribution is trivial, but how about more than one?
k nearest neighbor search, without brute force
k nearest neighbor search can be time consuming with brute force, how can we do better?
The fundamental problem of causal inference, part 2
Now we know what a good causal model look like, the next question is, how to build one?
The fundamental problem of causal inference, part 1
We all know that correlation does not imply causation. While we can observe correlations, how can we go about study causations?
Comparing two ranked lists
How similar is my top greatest movies of all time list to yours?
Hello World
Hello World, again!
2017
Quantum Nondemolition Measurements
One of the most well-known effects of quantum mechanics is the uncertainty principle. A direct consequence is that it that imposes a fundamental limit on the...
Normal mode
We are all quite familiar with a simple harmonic oscillator, such as a frictionless pendulum, or a mass-on-a-spring system. Since there is no friction (or en...
2016
I want to talk about light
What a poetic opening to your talk! This is what Sir John Pendry did when he gave a talk at the IIN Symposium few months back. He then went to describe light...
What I talk about when I talk about running
Pain is inevitable. Suffering is optional.