91
New
The latest blog posts to take off on Hacker News. 185 posts.
92
AI Explorables: big ideas in machine learning, simply explained
People + AI Research (PAIR) is a multidisciplinary team at Google that explores the human side of AI.
93
Probably Approximately Correct – A Formal Theory of Learning (2014)
In tackling machine learning (and computer science in general) we face some deep philosophical questions. Questions like, “What does it mean to learn?” and, “Can a computer learn?” and, “How do you define simplicity?” and, “Why does Occam’s Razor work? (Why do simple hypotheses do well at modelling reality?)” In a…
94
95
96
Ray: A Distributed Framework for Emerging AI Applications
Ray: A Distributed Framework for Emerging AI Applications Moritz, Nishihara, Wang et. al.
97
Reasonable Effectiveness of the Multiplicative Weights Update Algorithm (2017)
papad Hard to believe Sanjeev Arora and his coauthors consider it “a basic tool [that should be] taught to all algorithms students together with divide-and-conquer, dynamic programming, and random sampling.” Christos Papadimitriou calls it “so hard to believe that it has been discovered five times and forgotten.” It…
98
I do not agree with Github's use of copyrighted code as training for Copilot
I do not agree with GitHub's unauthorized and unlicensed use of copyrighted source code as training data for their ML-powered GitHub Copilot product. This product injects source code derived from copyrighted sources into the software of their customers without informing them of the license of the original source code…
99
Wav2vec Overview: Semi and Unsupervised Speech Recognition
A mid-depth cap with classic style for any occasion, this Nike Club Cap comes in smooth cotton twill that's been softly washed for casual, daily comfort. Soft and unstructured, the precurved bill makes for easy styling, and the adjustable back-strap lets you find the right fit. Nike Hat Size 6 1/2 6 5/8 6 3/4 6 7/8 7…
100
Alien Dreams: An Emerging Art Scene
Machine Learning at Berkeley is a student organization at UC Berkeley. Click to read ML@B Blog, by Machine Learning at Berkeley, a Substack publication with thousands of subscribers.
101
102
Fun and dystopia with AI-based code generation using GPT-J-6B
At the least, AI-generated code is much more readable than the average human’s.
103
FastText is the best text classification library for a quick baseline
Text classification is a very frequent use case for machine learning (ML) and natural language processing (NLP). It’s used for things like spam detection in emails, sentiment analysis for social media posts, or intent detection in chat bots. In this series I am going to compare several libraries that can be used to…
104
Advancing AI theory with first-principles understanding of deep neural networks
Deep neural networks have long been considered too complex to understand from first principles — but new research does just that, presenting a theoretical framework for DNNs.
105
1931: Kurt Gödel shows limits of math, logic, computing, AI
His famous 1931 paper identified the fundamental limits of math, logics, theorem proving, computing, AI.
106
Cuda.jl v3.3: union types, debug info, graph APIs
There have been several releases of CUDA.jl in the past couple of months, with many bugfixes and many exciting new features to improve GPU programming in Julia: CuArray now supports isbits Unions, CUDA.jl can emit debug info for use with NVIDIA tools, and changes to the compiler make it even easier to use the latest…
107
Gemini's "uselessness" is its killer feature
So it’s borderline useless for all but the simplest use cases, and even there it does a very subpar job. I’d like to have an inline image every now and then, or organize things in some multi-column layouts or the like, or a table. 19th century newspapers could do this, and if your technology can’t emulate typography…
108
Using aligned word vectors for instant translations with Python and Rust
How to use aligned word vectors from fastText to translate words from one language to another instantly.
109
GPT-J-6B: 6B JAX-Based Transformer
Summary: We have released GPT-J-6B, 6B JAX-based (Mesh) Transformer LM (Github).GPT-J-6B performs nearly on par with 6.7B GPT-3 (or Curie) on various zero-shot down-streaming tasks.You can try out …
110
Text Classification by Data Compression
Edit – I posted this on Hackernews and got some valuable feedback. Many brought up the fact that you should be able to reuse the internal state of the compressor instead of recompressing the training …
111
“AI promised to revolutionize radiology but so far its failing”
Geoffrey Hinton is a legendary computer scientist . . . Naturally, people paid attention when Hinton declared in 2016, “We should stop training radiologists now, it’s just completely obvious within five years deep learning is going to do better than radiologists.” The US Food and Drug Administration (FDA) approved the…
112
Yann LeCun Deep Learning Course 2021
Most of the lectures, labs, and notebooks are similar to the previous edition, nevertheless, some are brand new. I will try to make clear which is which.
113
Do wide and deep networks learn the same things?
Posted by Thao Nguyen, AI Resident, Google Research A common practice to improve a neural network’s performance and tailor it to available computat...
114
A Browsable Petascale Reconstruction of the Human Cortex
Posted by Tim Blakely, Software Engineer and Michał Januszewski, Research Scientist, Connectomics at Google In January 2020 we released the fly “he...
115
PID Controller: A Simple Control Loop Mechanism
A mid-depth cap with classic style for any occasion, this Nike Club Cap comes in smooth cotton twill that's been softly washed for casual, daily comfort. Soft and unstructured, the precurved bill makes for easy styling, and the adjustable back-strap lets you find the right fit. Nike Hat Size 6 1/2 6 5/8 6 3/4 6 7/8 7…
116
Potemkin Data Science (2020)
Huge swathes of people, in Europe and North America in particular, spend their entire working lives performing tasks they secretly believe…
117
118
High-performance speech recognition with no supervision at all
To enable speech recognition technology for many more languages spoken around the globe, Facebook AI is releasing wav2vec Unsupervised, a new method to train models with no supervision whatsoever. Wav2vec Unsupervised rivals the performance of the best supervised systems from just a few years ago.
119
MUM: A new AI milestone for understanding information
Our latest AI milestone, Multitask Unified Model, has the potential to take Google’s information understanding to a new level.
120
