The Innovation | Informatics

Spatial-Temporal Identity: A Simple yet Effective Baseline for Multivariate Time Series Forecasting A promising method nearly 200 stars, and used as baselines by nearly 100 papers. github.com/GestaltCogTeam… #timeseries #MachineLearning #deeplearning #artificalintelligence #bigdata

Spatial-Temporal-Decoupled Masked Pre-training for Spatiotemporal Forecasting arxiv.org/abs/2312.00516 #ChatGPT #Timeseries #datascience #artificalintelligence #Prediction #deeplearning #MachineLearning #STGNN #bigdata

UniST: A Prompt-Empowered Universal Model for Urban Spatio-Temporal Prediction dl.acm.org/doi/abs/10.114… #ChatGPT #Timeseries #datascience #urbancomputing #artificalintelligence #Prediction #deeplearning #MachineLearning #STGNN #bigdata @Tsinghua_Uni

Common occurrences of subsurface heatwaves and cold spells in ocean eddies nature.com/articles/s4158… #Science #heatwaves #timeseries #ocean

Chronos: Learning the Language of Time Series arxiv.org/abs/2403.07815 #ChatGPT #Timeseries #datascience #artificalintelligence #Prediction #deeplearning #MachineLearning #STGNN #bigdata @amazon

This's why we insist in whole-brain decoding. "behavioral signatures can be decoded from a much broader range of cortical areas than previously recognized" direct.mit.edu/imag/article/d…

Pre-training enhanced spatial-temporal graph neural network for multivariate time series forecasting dl.acm.org/doi/abs/10.114… #ChatGPT #Timeseries #datascience #artificalintelligence #Prediction #deeplearning #MachineLearning #STGNN #bigdata

BigST: Linear Complexity Spatio-Temporal Graph Neural Network for Traffic Forecasting on Large-Scale Road Networks dl.acm.org/doi/10.14778/3… #ChatGPT #Timeseries #datascience #artificalintelligence #Prediction #deeplearning #MachineLearning #STGNN #bigdata

Excited to share work from my internship at @AIatMeta! LLM devs often tweak decoding temperature: low for analytical tasks, and high for creative ones. Why not learn this from the data? Introducing the AdaptiveDecoder! (1/3)🧵

I trained a 2-arm robot using my Nintendo Switch (+ an action transformer) how I did it + my most interesting takeaways below👇🏼

Microsoft released LLM2CLIP: a CLIP model with longer context window for complex text inputs 🤯 TLDR; they replaced CLIP's text encoder with various LLMs fine-tuned on captioning, better top-k accuracy on retrieval 🔥 All models with Apache 2.0 license on @huggingface 😍

A new promising “A Fair and Scalable Time Series Forecasting Benchmark and Toolkit” BasicTS. #timeseries #forecasting #opensource

Forget traditional methods like SMOTE—Conformal Prediction offers significant improvements, even on challenging datasets with imbalanced data, label noise, or both. 🚀 #conformalprediction #MachineLearning #DataScience #Innovation

Microsoft released a neat Python library to run LLM-powered multi-agent simulations of people with personalities, interests, and goals.

How to write an excellent Review article @natrevbioeng @Nature #PhD #PhDlife #academics #AcademicChatter #AcademicTwitter #Postdocs #Research #Science

New article: "The geometry of data: the missing metric tensor and the Stein score" (blog.christianperone.com/2024/11/the-ge…). I show how you can derive a (efficient to compute) data manifold metric tensor with the Stein score alone ! Deep connections to diffusion, score-based models and physics.

New in The Innovation Geoscience! Larger increase in future global terrestrial water availability than projected by CMIP6 models. Wu et al. employ outputs from 20 state-of-the-art ESMs from the Coupled Model Intercomparison Project Phase 6, together with multiple remote sensing and ground-based precipitation and ET datasets to derive new projections of ΔPME under shared socioeconomic pathway 2-4.5 and 5-8.5 scenarios using the EC approach. Read more @Innov_Geosci the-innovation.org/article/doi/10… #geoscience #Research #climate

A new Science study presents “Evo”—a machine learning model capable of decoding and designing DNA, RNA, and protein sequences, from molecular to genome scale, with unparalleled accuracy. Evo’s ability to predict, generate, and engineer entire genomic sequences could change the way synthetic biology is done. Learn more in this week's issue: bit.ly/3OsmUPr

#Multiomics integration unravels #genotype‐#microbiome interactions shaping the conjunctival transcriptome #iMetaOmics onlinelibrary.wiley.com/doi/10.1002/im… #bioinformatics #DataScience #MachineLearning @wileymicrobio @WileyBiomedical @WileyHealth

Artificial intelligence for geoscience: Progress, challenges, and perspectives cell.com/the-innovation… #ArtificialIntelligence #MachineLearning #Geoscience #Science #Research #Trending #PERSPECTIVE #AI #Technology