Examining the design of interactive articles by synthesizing theory from disciplines such as education, journalism, and visualization.
With diverse environments, we can analyze, diagnose and edit deep reinforcement learning models using attribution.
Neural networks naturally learn many transformed copies of the same feature, connected by symmetric weights.
A family of early-vision neurons reacting to directional transitions from high to low spatial frequency.
We present techniques for visualizing, contextualizing, and understanding neural network weights.
Neural Cellular Automata learn to generate textures, exhibiting surprising properties.
When a neural network layer is divided into multiple branches, neurons self-organize into coherent groupings.
Weights in the final layer of common visual models appear as horizontal bands. We investigate how and why.
Reprogramming Neural CA to exhibit novel behaviour, using adversarial attacks.
What components are needed for building learning algorithms that leverage the structure and properties of graphs?
Understanding the building blocks and design choices of graph neural networks.
The Opportunity in Arabic Digital Publishing The Arabic-speaking world has over 400 million native speakers, yet the Kindle catalog for Arabic content remains surprisingly thin — especially in the sel...
Claude Code's Hooks system lets you run scripts automatically before or after every tool call — code writes, bash commands, file edits. Configure them once, and every interaction gets automatic qualit...
Technical debt accumulates silently. A function that started at 20 lines grows to 120. A variable named x makes sense to the original author at 2am but confuses everyone else six months later. Duplica...
Every week, security researchers publish new CVEs. Every week, production systems get breached because someone didn't update a dependency in time. The gap between "vulnerability disclosed" and "team p...
