Digests » 160


Machine learning for voice made easy

Why doesn’t more software use voice? Machine learning for voice is hard, and existing solutions are clunky and rigid. So we open-sourced libraries for Node, Python, Android, iOS, & React Native, and built a no-code web tool to make custom wake words, speech recognizers, and AI voices—for all devs!

this week's favorite

A concrete introduction to probability

Laplace nailed it. To untangle a probability problem, all you have to do is define exactly what the cases are, and careful count the favorable and total cases. Let's be clear on our vocabulary words.

Exploratory data analysis in Python

As a common saying goes, it’s often not the drawing that’s the hard part; instead, it’s deciding what to draw that gets the best of most. Fortunately, once you’ve chosen, the rest isn’t as hard as it might seem. The same is the case in data science, and the phase where you’re *deciding* is what we refer to as EDA or Exploratory Data Analysis.

Building stronger semantic understanding into text game reinforcement learning agents

In this blog post, we share two papers that explore reinforcement learning methods to improve semantic understanding in text agents, a key process by which AI understands and reacts to text-based input. We’re also releasing source code for these agents to encourage the community to continue to improve semantic understanding in text-based games.

When Vision Transformers outperform ResNets without pretraining or strong data augmentations

Vision Transformers (ViTs) and MLPs signal further efforts on replacing hand-wired features or inductive biases with general-purpose neural architectures. Existing works empower the models by massive data, such as large-scale pretraining and/or repeated strong data augmentations, and still report optimization-related problems (e.g., sensitivity to initialization and learning rate).

How can generative adversarial networks learn real-life distributions

A Generative adversarial network, or GAN, is one of the most powerful machine learning models proposed by Goodfellow et al. for learning to generate samples from complicated real-world distributions. GANs have sparked millions of applications, ranging from generating realistic images or cartoon characters to text-to-image translations. Turing award laureate Yann LeCun called GANs “the most interesting idea in the last 10 years in ML.”