Machine Learning

Distill Takes a Break: What the Pause Means for the ML Visualization Community

After five years of publishing, Distill is taking a break.

Jul 02, 2021 1,012 views

Smart Weight Banding: A Practical Guide to Tiered Weight Classification

Final-layer weights in common visual models form distinct horizontal band patterns — here's what's driving that structure and why it matters.

Apr 08, 2021 651 views

Smart Frequency Detection: Bridging High and Low Bands for Modern Applications

Neurons in the early visual system selectively respond to directional spatial frequency transitions — specifically, shifts from high to low frequency — forming a functionally distinct neural population at the front end of visual processing.

Jan 27, 2021 638 views

How Neural Networks Develop Equivariance on Their Own

Symmetry in neural network weights causes redundant feature representations to emerge naturally during training.

Dec 08, 2020 978 views

Mastering Reinforcement Learning for Computer Vision: A Practical Guide

Attribution techniques unlock the ability to analyze, diagnose, and refine deep reinforcement learning models across varied training environments.

Nov 17, 2020 502 views

How InceptionV1 Processes Images in Its Early Vision Layers

A structured taxonomy of the neurons spanning the first five layers of InceptionV1, grouping them by shared functional characteristics to map how the network builds visual representations from low-level features upward.

Apr 01, 2020 661 views

Exploring Neural Networks in Motion: How the Grand Tour Makes High-Dimensional Data Visible

Linear dimensionality reduction reveals the hidden geometry of neural network dynamics, making complex, high-dimensional behavior interpretable through clean, low-dimensional visualizations.

Mar 16, 2020 439 views

Getting Started with Circuits: A Beginner's Guide to How Electronics Work

Neural network weights encode more than learned parameters — they reflect the underlying algorithms shaped by training. Analyzing the connectivity patterns between neurons can surface these embedded computational structures, offering a clearer window into how models actually process information.

Mar 10, 2020 1,007 views

How Value Learning Actually Works: A Paths-Based Perspective

Temporal Difference Learning sidesteps the inefficiency of waiting for full episode returns by bootstrapping value estimates from intermediate steps. This piece examines how TD methods blend Monte Carlo sampling with dynamic programming — updating predictions mid-trajectory rather than post-hoc — to extract more signal from each transition. The result is faster convergence and stronger sample efficiency, particularly in environments where episodes are long or rewards are sparse.

Sep 30, 2019 1,032 views

Why "Adversarial Examples Are Features, Not Bugs" Should Change How We Define Model Robustness

Ilyas et al. (2019)'s central hypothesis is a specific instance of a broader, well-established principle in the distributional shift robustness literature.

Aug 06, 2019 399 views

Rethinking Adversarial Examples: How Neural Style Transfer Becomes More Robust by Design

Adversarial robustness unlocks neural style transfer beyond VGG — an experimental finding showing that robust training enables style transfer to generalize across non-VGG architectures.

Aug 06, 2019 742 views

Why Adversarial Examples Are Features, Not Flaws: What Mislabeled Data Reveals About How Neural Networks Really Learn

Section 3.2 of Ilyas et al. (2019) demonstrates that a model trained exclusively on adversarial examples achieves non-trivial generalization on the original test set. This paper shows that these experiments represent a specific instance of a broader phenomenon: adversarial vulnerability can emerge from features that are genuinely predictive yet brittle, rather than from model artifacts or noise. The finding reframes adversarial robustness not as a bug to be patched, but as a structural property of the data distribution itself.

Aug 06, 2019 575 views

Unsolved Challenges in Generative Adversarial Networks That Still Define the Field

Open questions in GAN research: what the field still hasn't figured out.

Apr 09, 2019 935 views

How RNNs Actually Memorize Sequences — and What It Looks Like

Gradient magnitude inspection reveals how recurrent units weight short-term versus long-term context during processing — a practical diagnostic for understanding temporal dependencies in sequence models.

Mar 25, 2019 670 views

What's New in Distill: 2018 Update Highlights

I can't really rewrite that — "An Update from the Editorial Team" doesn't contain any actual content or facts to work with. It's just a title. Share the full article text or summary and I'll get it done.

Aug 14, 2018 1,001 views

How AI Is Expanding the Boundaries of Human Intelligence

Designing interfaces around a model's internal representations unlocks new cognitive tools — giving people more direct ways to explore, interrogate, and reason with machine learning systems.

Dec 04, 2017 1,031 views

What Actually Makes Momentum Work — and How to Use It

Momentum-based optimization is commonly visualized as a ball rolling downhill — a useful analogy, but one that only scratches the surface of what's actually happening mathematically and dynamically during gradient descent.

Apr 04, 2017 1,066 views

Teaching Machines to Write: Inside Neural Network Handwriting Experiments

Explore a collection of interactive visualizations built on a generative handwriting model — ranging from playful experiments to practical demonstrations of how the model behaves.

Dec 06, 2016 952 views