Blog Post Guidelines

Link to Blog: https://2020blogfor.github.io/

Demonstration Template Post: https://2020blogfor.github.io/posts/2017/08/statistical_whitening/

Submission Instructions: Submit images and markdown files to Hao Liu (hliu3@caltech.edu). [Template files]

Due: April 24th (end of day).

Goals of the Blog Post

The goal of the blog post is to get students to demonstrate & practice the ability to read and understand technical concepts in research papers. The blog post need not necessarily be tied to the final research project (although it usually is a good idea).

Writing a Good Blog Post

As shown in the demonstration template post, a good blog post should have the following features:

• A high-level introduction of the topic and its surrounding context, including motivating its importance.

• A set of simple examples (including figures, pseudo-code, or both) that clearly illustrates the key intuition.

• A discussion of important implications, including (depending on the topic):

  • Theoretical/formulaic derivations

  • Key extensions that make the idea practically relevant

  • Connections to other topics

  • More advanced numerical/empirical results

• A summary of some follow-up work that builds upon this key idea.

  • E.g., later papers that build upon the core idea

The target length is a 10-15 minute read. (The markdown language provides an estimate of the reading time.)

Recommended Starter Papers for topics (you can choose something else if you want):

• Gaussian Process Optimization

• SMAC

• SATzilla

• Hyperband

• Learning to Learn by Gradient Descent by Gradient Descent

• Model-Agnostic Meta Learning

• Learning Search for Branch-and-Bound

• OptNet

• Learning Combinatorial Optimization over Graphs

• Large Neighborhood Search