Tag
Researchers show that DFT-relaxed structural distortions cause machine-learning screens to misclassify hidden altermagnets as antiferromagnets, and propose a tolerance-aware symmetry pipeline that discovers six new altermagnet materials including Ba₂FeMoO₆ and Sr₂FeWO₆, with spin splitting of 85–405 meV at zero net magnetization.
An article discussing how artificial intelligence is poised to transform materials science, accelerating the discovery and design of new materials through ML-driven approaches.
This arXiv paper fuses SEC financial data with EPA greenhouse gas registries, using gradient boosting and Mondrian Conformal Prediction to build an algorithmic baseline for corporate emissions and quantify greenwashing via a Conformal-Weighted Continuous Divergence (CWCD) metric, finding that emissions divergence negatively predicts market valuation and profitability.
This paper proposes HyMLRaman, a hybrid framework combining EfficientNet-B3 spectral embeddings with classical classifiers (notably SVM) and DDPM-based generative feature augmentation to identify six pharmaceutical compounds via Raman spectroscopy, achieving 96.31% accuracy.
A University of Bern team developed a machine-learning algorithm trained on tens of thousands of simulated planetary systems that identified 44 known star systems which may host undiscovered Earth-like planets, with up to 99% precision on simulated data but still awaiting telescope confirmation.
An analysis of four years of AI progress across seven capability areas, with results adjusted for changes and drift in the underlying benchmarks to better reflect true model improvement.
antirez argues that calling classifiers "decision models" misrepresents machine learning, since such models (whether small LLMs using prefill logits or BERT-like encoders) make far fewer decisions than an LLM that processes information through chain-of-thought reasoning.
MIT News profiles Cathy Wu, an associate professor in CEE and IDSS, whose research applies machine learning and reinforcement learning to design more reliable and efficient transportation systems, and whose early RL-based traffic analysis research went viral.
A former Google DeepMind researcher reflects on AlphaGo's Move 37 and argues that today's LLMs still lack genuine, auditable reasoning, calling for new architectures inspired by AlphaGo's search-and-reason approach in a new MIT Technology Review piece.
A profile of Gruber, a marine biologist and founder of Project CETI (Cetacean Translation Initiative), who is using artificial intelligence and machine learning to decode sperm whale communication.
A discussion post raising concerns that future AGI systems, trained on internet data full of racist content, may absorb and amplify subtle biases against minorities, citing past incidents like Tay, Grok, and Amazon's biased hiring AI.
Mathematician Terence Tao reflects on modern AI, arguing that current neural network systems feel like clever tricks rather than genuine intelligence, despite solving real problems, noting we still lack an elusive 'intelligent way of thinking' in tools that achieve our goals.
This paper characterizes when regression is possible under sample selection biases that depend on both covariates and outcomes, showing the regression function can be identified even when the selection filter is not. It provides finite-sample estimation guarantees, oracle-efficient algorithms, and the first general-purpose estimation method for this class of selection problems.
This paper proposes BaryFM, a flow matching model that learns to approximate the entire Wasserstein simplex — the full family of Wasserstein barycenters over the simplex of weight vectors — rather than a fixed-weight barycenter. BaryFM achieves the best average rank among 15 methods across 10 domain adaptation benchmarks and is validated on tasks including generalization, Bayesian posterior aggregation, and algorithmic fairness.
本文研究通过 e-scooter 车载传感器（IMU、油门/刹车位置）数据结合机器学习检测骑行者酒精中毒，利用排列熵与标准差提取运动学特征，逻辑回归分类器在留一被试交叉验证下达到 85% 准确率和 0.94 的 AuROC。
A new paper introduces CAGE, a framework combining generative models, adversarial discrimination, and conformal prediction to weight time series ensemble forecasts, showing improved robustness against outliers and noise on milk collection and monkeypox datasets.
A new moving-horizon approximate branch-and-reduce method for training near-optimal deep classification trees on large-scale datasets with continuous features, achieving better accuracy than heuristic baselines and far greater scalability than global optimal solvers.
A doctoral dissertation-based paper exploring machine learning and deep learning methods (including 1D-CNN) for predicting freight travel times in complex multimodal supply chains, aiming to improve logistics planning, ETA accuracy, and lead time management.
A detailed mathematical explainer of the softmax function, its probabilistic interpretation in multiclass classification, and a full derivation of its Jacobian/derivative using the quotient rule, including the gradient of a softmax-based loss.
Sony is bringing Quick Spectral Super Resolution (QSSR), an AI graphics upscaling technology born from its Project Amethyst work with AMD, to the standard PS5. The first titles to adopt it are Marvel's Wolverine and Ghost of Yōtei, which are getting patches today.