Introduction Neural networks can learn to classify images more accurately than any system humans directly design. This raises a natural…
Read More
AI Safety Needs Social Scientists
The goal of long-term artificial intelligence (AI) safety is to ensure that advanced AI systems are reliably aligned with human…
Read More
Distill Update 2018
A little over a year ago, we formally launched Distill as an open-access scientific journal.Distill operated informally for several months…
Read More
Differentiable Image Parameterizations
Neural networks trained to classify images have a remarkable — and surprising! — capacity to generate images. Techniques such as DeepDream , style transfer,…
Read More
Feature-wise transformations
Many real-world problems require integrating multiple sources of information. Sometimes these problems involve multiple, distinct modalities of information — vision, language, audio,…
Read More
The Building Blocks of Interpretability
With the growing success of neural networks, there is a corresponding need to be able to explain their decisions — including building…
Read More
Using Artificial Intelligence to Augment Human Intelligence
What are computers for? Historically, different answers to this question – that is, different visions of computing – have helped…
Read More
Sequence Modeling with CTC
Introduction Consider speech recognition. We have a dataset of audio clips and corresponding transcripts. Unfortunately, we don’t know how the…
Read More
Feature Visualization
There is a growing sense that neural networks need to be interpretable to humans. The field of neural network interpretability…
Read More
Why Momentum Really Works
Step-size α = 0.02 Momentum β = 0.99 We often think of Momentum as a means of dampening oscillations and…
Read More
