ResearchRLHF alignment
RVPO: Risk-Sensitive Alignment via Variance Regularization
Current critic-less RLHF methods aggregate multi-objective rewards via an arithmetic mean, leaving them vulnerable to constraint neglect:… machinelearning.apple.com
🔬 AI Research
Academic papers, research breakthroughs, and scientific advances in machine learning and AI.
90 articles
Researchreinforcement learning
New work by Turing Award winner Sutton: Solve a major flaw in streaming reinforcement learning with a 1967 formula
At the end of 2024, a paper titled "Streaming Deep Reinforcement Learning Finally Works" (arXiv:2410.14606) sparked extensive discussions in the academic... eu.36kr.com
Researchgenerative models / normalizing flows
Normalizing Flows with Iterative Denoising
Normalizing Flows (NFs) are a classical family of likelihood-based methods that have received revived attention. Recent efforts such as… machinelearning.apple.com
ResearchML in clinical trials
Revolutionizing clinical trials with machine learning
image: Design Framework for Adaptive Randomization with Interim Analyses. view more. Credit: Y.Park and S.Nycklemoe, “MARGO: Machine Learning-Assisted... eurekalert.org
ResearchML in medical research / Parkinson's subtypes
Machine learning reveals two main Parkinson's types and five subgroups
A new study led by researchers from VIB and KU Leuven shows that Parkinson's disease can be divided into distinct subtypes, helping explain why a single... medicalxpress.com
ResearchML in medical research / Parkinson's subtypes
Machine learning identifies five distinct Parkinson’s disease subtypes
A new study from VIB and KU Leuven has revealed that Parkinson's disease comprises five distinct molecular subtypes, each requiring tailored therapeutic... drugtargetreview.com
Researchquantum machine learning - noise benefits
Quantum Machine Learning Benefits From Added Noise, Study Reveals
Operational noise, typically detrimental to quantum systems, unexpectedly improves the performance of newly constructed quantum kernels. quantumzeitgeist.com
NewsAI research integrity/retraction
Influential study touting ChatGPT in education retracted over red flags
A study that claimed OpenAI's ChatGPT can positively impact student learning has been retracted nearly one year after publication. arstechnica.com
NewsAI research digest/self-improving systems
Import AI 455: AI systems are about to start building themselves.
Welcome to Import AI, a newsletter about AI research. Import AI runs on arXiv and feedback from readers. If you'd like to support this, please subscribe. importai.substack.com
ResearchAI for scientific computing/PDEs
Penn engineers use AI to solve some of science’s most difficult math problems
Penn engineers built an AI math tool that solves inverse PDEs faster and more reliably, even with noisy data. thebrighterside.news
ResearchAI for inverse PDE problems
AI method tackles one of science's hardest math problems
Penn Engineers develop new framework for inverse partial differential equation learning, which could help researchers infer hidden dynamics in systems... eurekalert.org
ResearchML in healthcare/genomics
Machine learning in prediction and classification of type 1 diabetes
Genetic prediction of type 1 diabetes is one of the most successful for complex traits. A machine learning approach now improves this further and discovers... nature.com
ResearchAI fairness / debiasing
Solving the “Whac-a-mole dilemma”: A smarter way to debias AI vision models
A new debiasing approach called WRING resolves the "Whac-a-Mole dilemma" of existing debiasing approaches that can create or amplify existing biases. news.mit.edu
ResearchLLM reasoning / diffusion models
LaDiR: Latent Diffusion Enhances LLMs for Text Reasoning
Large Language Models (LLMs) demonstrate their reasoning ability through chain-of-thought (CoT) generation. However, LLM's autoregressive… machinelearning.apple.com
Researchrepresentation learning in healthcare
Advancing Multi-Institutional EHR Studies via Representation Learning
In a groundbreaking development poised to revolutionize medical research across borders, a team of scientists led by Zhou, D., Tong, H., and Wang,... bioengineer.org
Researchbiological neural networks
Researchers train living rat neurons to perform…
Japanese researchers trained cultured rat cortical neurons to autonomously generate complex temporal signals using a real-time machine learning… inkl.com
Researchfoundation models, biological aging prediction
Meet MaxToki: The AI That Predicts How Your Cells Age — and What to Do About It
Researchers at Gladstone Institutes have built a temporal foundation model that forecasts cell state trajectories across the human lifespan — and... marktechpost.com
LaunchAI for astronomy
Carnegie Mellon Launches New Effort To Advance AI-Driven Astronomy
At Carnegie Mellon University, a new initiative will bring together experts in AI, statistics and astrophysics to accelerate that shift. cmu.edu
Researchbiological neural networks
Wetware AI: Living Brain Cells Trained to Run Chaos Math
Can living neurons replace AI? A new study shows that biological neural networks (BNNs) can be trained to perform reservoir computing. neurosciencenews.com
Researchbiological computing and neural networks
Biological neural networks may serve as viable alternatives to machine learning models
A research team at Tohoku University and Future University Hakodate has demonstrated that living biological neurons can be trained to perform a supervised... news-medical.net