We study partial-feedback online learning, where each instance admits a set of correct labels, but the learner only observes one correct label per round; any prediction within the correct set is counted as correct. This model captures settings such as language generation, where multiple responses may be valid but data provide only a single reference. We give a near-complete characterization of minimax regret for both deterministic and randomized learners in the set-realizable regime, i.e., in the regime where sublinear regret is generally attainable. For deterministic learners, we introduce the Partial-Feedback Littlestone dimension (PFLdim) and show it precisely governs learnability and minimax regret; technically, PFLdim cannot be defined via the standard version space, requiring a new collection version space viewpoint and an auxiliary dimension used only in the proof. We further develop the Partial-Feedback Measure Shattering dimension (PMSdim) to obtain tight bounds for randomized learners. We identify broad conditions ensuring inseparability between deterministic and randomized learnability (e.g., finite Helly number or nested-inclusion label structure), and extend the argument to set-valued online learning, resolving an open question of Raman et al. [2024b]. Finally, we show a sharp separation from weaker realistic and agnostic variants: outside set realizability, the problem can become information-theoretically intractable, with linear regret possible even for $|H|=2$. This highlights the need for fundamentally new, noise-sensitive complexity measures to meaningfully characterize learnability beyond set realizability.

Learning causal structures from observational data remains a fundamental yet computationally intensive task, particularly in high-dimensional settings where existing methods face challenges such as the super-exponential growth of the search space and increasing computational demands. To address this, we introduce VISTA (Voting-based Integration of Subgraph Topologies for Acyclicity), a modular framework that decomposes the global causal structure learning problem into local subgraphs based on Markov Blankets. The global integration is achieved through a weighted voting mechanism that penalizes low-support edges via exponential decay, filters unreliable ones with an adaptive threshold, and ensures acyclicity using a Feedback Arc Set (FAS) algorithm. The framework is model-agnostic, imposing no assumptions on the inductive biases of base learners, is compatible with arbitrary data settings without requiring specific structural forms, and fully supports parallelization. We also theoretically establish finite-sample error bounds for VISTA, and prove its asymptotic consistency under mild conditions. Extensive experiments on both synthetic and real datasets consistently demonstrate the effectiveness of VISTA, yielding notable improvements in both accuracy and efficiency over a wide range of base learners.

Go to a small liberal-arts college if you can. I n the waning heat of last summer, freshly back in my office at a major research university, I found myself considering the higher-education hellscape that had lately descended upon the nation. I'd spent months reporting on the Trump administration's attacks on universities for, speaking with dozens of administrators, faculty, and students about the billions of dollars in cuts to public funding for research and the resulting collapse of " college life ."At Initially, I surveyed the situation from the safe distance of a journalist who happens to also be a career professor and university administrator. I saw myself as an envoy between America's college campuses and its citizens, telling the stories of the people whose lives had been shattered by these transformations. By the summer, though, that safe distance had collapsed back on me.

The government has launched a series of free AI training courses designed to help people learn how to use the technology at work. The online lessons give advice on things such as how to prompt chatbots or use them to assist with admin tasks. Many of the courses are free, with others subsidised, and the government aims to reach 10 million workers by 2030 - calling it the most ambitious training scheme since the launch of the Open University in 1971. But the Institute for Public Policy Research (IPPR) has warned workers will need to know more than just how to prompt a chatbot as the workforce adapts to the growth of AI. Skills for the age of AI can't be reduced to short technical courses alone, said Roa Powell, senior research fellow at the IPPR.

We study how to fine-tune LLMs using user-edit deployment data consisting of a set of context, an agent's response, and user edits. This deployment data is naturally generated by users in applications such as LLMs-based writing assistants and coding agents. The _natural_ origin of user edits makes it a desired source for adapting and personalizing LLMs. In this setup, there emerges a unification of various feedback types namely preferences, supervised labels, and cost that are typically studied separately in the literature. In this paper, we initiate the theoretical investigation of learning from user edits. We first derive bounds for learning algorithms that learn from each of these feedback types. We prove that these algorithms have different trade-offs depending upon the user, data distribution, and model class. We then propose a simple ensembling procedure to jointly learn from these feedback types. On two domains adapted from Gao et al. 2024, we show our ensembling procedure outperforms these methods that learn from individual feedback. Further, we show that our proposed procedure can robustly adapt to different user-edit distributions at test time.

Chemist Omar Yaghi invented materials called MOFs, a few grams of which have the surface area of a football field. In school, we learn about the Stone Age, the Bronze Age - and we are currently in a silicon age characterised by computers and phones. What might define the next age? Omar Yaghi at the University of California, Berkeley, thinks a family of materials he helped pioneer in the 1990s has a good shot. They are metal-organic frameworks (MOFs), and working out how to make them earned him a share of the 2025 Nobel prize in chemistry .

This article presents the full, original record of the 2024 Joint Statistical Meetings (JSM) town hall, "Statistics in the Age of AI," which convened leading statisticians to discuss how the field is evolving in response to advances in artificial intelligence, foundation models, large-scale empirical modeling, and data-intensive infrastructures. The town hall was structured around open panel discussion and extensive audience Q&A, with the aim of eliciting candid, experience-driven perspectives rather than formal presentations or prepared statements. This document preserves the extended exchanges among panelists and audience members, with minimal editorial intervention, and organizes the conversation around five recurring questions concerning disciplinary culture and practices, data curation and "data work," engagement with modern empirical modeling, training for large-scale AI applications, and partnerships with key AI stakeholders. By providing an archival record of this discussion, the preprint aims to support transparency, community reflection, and ongoing dialogue about the evolving role of statistics in the data- and AI-centric future.

The grade of membership model is a flexible latent variable model for analyzing multivariate categorical data through individual-level mixed membership scores. In many modern applications, auxiliary covariates are collected alongside responses and encode information about the same latent structure. Traditional approaches to incorporating such covariates typically rely on fully specified joint likelihoods, which are computationally intensive and sensitive to misspecification. We introduce a covariate-assisted grade of membership model that integrates response and covariate information by exploiting their shared low-rank simplex geometry, rather than modeling their joint distribution. We propose a likelihood-free spectral estimation procedure that combines heterogeneous data sources through a balance parameter controlling their relative contribution. To accommodate high-dimensional and heteroskedastic noise, we employ heteroskedastic principal component analysis before performing simplex-based geometric recovery. Our theoretical analysis establishes weaker identifiability conditions than those required in the covariate-free model, and further derives finite-sample, entrywise error bounds for both mixed membership scores and item parameters. These results demonstrate that auxiliary covariates can provably improve latent structure recovery, yielding faster convergence rates in high-dimensional regimes. Simulation studies and an application to educational assessment data illustrate the computational efficiency, statistical accuracy, and interpretability gains of the proposed method. The code for reproducing these results is open-source and available at \texttt{https://github.com/Toby-X/Covariate-Assisted-GoM}

An exclusive conversation with Kevin Weil, head of OpenAI for Science, a new in-house team that wants to make scientists more productive. In the three years since ChatGPT's explosive debut, OpenAI's technology has upended a remarkable range of everyday activities at home, at work, in schools--anywhere people have a browser open or a phone out, which is everywhere. Now OpenAI is making an explicit play for scientists. In October, the firm announced that it had launched a whole new team, called OpenAI for Science, dedicated to exploring how its large language models could help scientists and tweaking its tools to support them. The last couple of months have seen a slew of social media posts and academic publications in which mathematicians, physicists, biologists, and others have described how LLMs (and OpenAI's GPT-5 in particular) have helped them make a discovery or nudged them toward a solution they might otherwise have missed. In part, OpenAI for Science was set up to engage with this community.

In the digital age, sound is proving to be the greatest connector of all, says Erik Vaveris, vice president of product management and CMO at Shure, and Brian Scholl, director of the Perception and Cognition Laboratory at Yale University. In an era where business, education, and even casual conversations occur via screens, sound has become a differentiating factor. We obsess over lighting, camera angles, and virtual backgrounds, but how we sound can be just as critical to credibility, trust, and connection. Both see audio as more than a technical layer: It's a human factor shaping how people perceive intelligence, trustworthiness, and authority in virtual settings. If you're willing to take a little bit of time with your audio set up, you can really get across the full power of your message and the full power of who you are to your peers, to your employees, your boss, your suppliers, and of course, your customers, says Vaveris. Scholl's research shows that poor audio quality can make a speaker seem less persuasive, less hireable, and even less credible. We know that [poor] sound doesn't reflect the people themselves, but we really just can't stop ourselves from having those impressions, says Scholl. We all understand intuitively that if we're having difficulty being understood while we're talking, then that's bad. But we sort of think that as long as you can make out the words I'm saying, then that's probably all fine. And this research showed in a somewhat surprising way, to a surprising degree, that this is not so. For organizations navigating hybrid work, training, and marketing, the stakes have become high. Vaveris points out that the pandemic was a watershed moment for audio technology. As classrooms, boardrooms, and conferences shifted online almost overnight, demand accelerated for advanced noise suppression, echo cancellation, and AI-driven processing tools that make meetings more seamless. Today, machine learning algorithms can strip away keyboard clicks or reverberation and isolate a speaker's voice in noisy environments. That clarity underpins the accuracy of AI meeting assistants that can step in to transcribe, summarize, and analyze discussions. The implications across industries are rippling. It empowers executives and creators alike to produce broadcast-quality content from the comfort of their home office. And it offers companies new ways to build credibility with customers and employees without the costly overhead of traditional production.