Transformers and Tabular Data. After transformers took over NLP, language, speech and vision, now they are aiming to solve ML tasks with tabular data.
You may forget VAEs and GANs. Transformers is all you need for generating tabular data. Check out Language Models are Realistic Tabular Data Generators (paper & code.) The team claims SotA results across real-world datasets.
A team @MITCSAIL just published a technique that outperforms prior deep-learning-based tabular classification methods. And is also on par with gradient-boosted trees approaches. See: TabLLM: Few-shot Classification of Tabular Data with Large Language Models.
According to researchers @UniofFriburg, this may revolutionise ML. A new tabular data classification method that uses TabPFN, a transformer that solves small tabular classification in 1 sec, and yields SOTA performance (better than hyperparameter-optimised gradient boosting.)
A while ago, Julien @Hugginface published a nice video on how to use TabTransformer for supervised learning with tabular data.
I leave you with this videocast on how transformers are used with tabular data @CapitalOne.
Have a nice week.
Getting Started in the World of Stable Diffusion
Standford VIP Cheatsheets for AI, ML & DL
How We Use Data Science @FC Barcelona
Bayesball: Bayesian Analytics for Pro Baseball Batters
GNNs for Product Recommendations @AmazonScience
DL for Search Rankings @Etsy
Notes from Standford Intro to Deep Generative Models
[Free] Statistical Rethinking - A Bayesian Course, 2022
All The Math You Need for ML (pdf, 2190 pages)
Awesome Diffusion Models
Share Data Machina with friends
[Tutorials] ML Model Interpretability for PyTorch
[iPynb] Bayesian Interrupted Time-series Analysis
Out-of-Distribution Detection via Embeddings or Predictions
MLOps with vetiver in R & Python: Q&A
[Free course] DL with R
DL for Medical Images Processing
MIT TinyML & Efficient DL
CMU Advanced NLP: Intro to Prompt Engineering
Full Stack DL, 2022
Graph NNs for NLP: A Survey
Gaussian-Bernoulli RBMs Without Tears
[Deepmind] Solving Reasoning Tasks with a Slot Transformer
Contrasting Ridge Regression & The Lasso
Miniselect: Practical and Generic Selection Algos
Generating Chess Puzzles with Genetic Algos
Deep Whole Body: Unified Robotic Motion & Manipulation
Natural Robotics Contest, Winner 2022
Quadrupeds Inspecting Construction Site @Gatwick Airport
Visualising US Midterm Elections Forecasts
Can the World Feed Itself Sustainably?
Interactive: City Accessibility Maps
Why Typescript for Real-time Data Transformations
An Engineer's Guide to Data Contracts
Orchestrating Data/ML Workflows at Scale @Netflix
Kudos - AI for Wallet Cards Recommendations
SwingVision - AI for Tennis Stats & Line Calling
Uniphore - Conversational AI for Call Centers
Massively Multi-domain Language Modeling Dataset
A Large-Scale, Multilingual, Speech-to-Speech Translations Corpus
Automatically Find & Fix Errors in your ML Datasets
Tips? Suggestions? Feedback? email Carlos
Curated by @ds_ldn in the middle of the night.