Abstract
Modern search systems play a crucial role in facilitating information acquisition. Traditional search engines typically rely on a cascaded architecture, where results are retrieved through recall, pre-ranking, and ranking stages. The complexity of designing and maintaining multiple modules makes it difficult to achieve holistic performance gains. Recent advances in generative recommendation have motivated the exploration of unified generative search as an alternative. However, existing approaches are not genuinely end-to-end: they typically train an item encoder to tokenize candidates first and then optimize a generator separately, leading to objective inconsistency and limited generalization. To address these limitations, we propose UniSearch, a unified generative search framework for Kuaishou Search. UniSearch replaces the cascaded pipeline with an end-to-end architecture that integrates a Search Generator and a Video Encoder. The Generator produces semantic identifiers of relevant items given a user query, while the Video Encoder learns latent item embeddings and provides their tokenized representations. A unified training framework jointly optimizes both components, enabling mutual enhancement and improving representation quality and generation accuracy. Furthermore, we introduce Search Preference Optimization (SPO), which leverages a reward model and real user feedback to better align generation with user preferences. Extensive experiments on industrial-scale datasets, together with online A/B testing in both short-video and live search scenarios, demonstrate the strong effectiveness and deployment potential of UniSearch. Notably, its deployment in live search yields the largest single-experiment improvement in recent years of our product's history, highlighting its practical value for real-world applications.

AI Insights

• GRAM introduces a novel alignment mechanism that bridges generated candidates and query semantics, boosting retrieval accuracy.
• The generative component of GRAM produces candidate documents directly from queries, eliminating the need for pre‑built indexes.
• Alignment is achieved via a learned similarity scorer that reorders candidates, outperforming traditional BM25 baselines on TREC datasets.
• Evaluation on multiple benchmarks (MS MARCO, Natural Questions) shows a 5–7 % MAP lift over state‑of‑the‑art generative models.
• Training GRAM requires GPU clusters; inference latency is ~50 ms per query, limiting real‑time deployment without optimization.
• Recommended reading: “Listwise Generative Retrieval Models via a Sequential Learning Process” (2024) for advanced training strategies.
• Core definition: Generative Retrieval – a retrieval paradigm that synthesizes candidate documents conditioned on the query.

Abstract
We regularly encounter information on novel, emerging topics for which the body of knowledge is still evolving, which can be linked, for instance, to current events. A primary way to learn more about such topics is through web search. However, information on emerging topics is sparse and evolves dynamically as knowledge grows, making it uncertain and variable in quality and trustworthiness and prone to deliberate or accidental manipulation, misinformation, and bias. In this paper, we outline a research vision towards search systems and interfaces that support effective knowledge acquisition, awareness of the dynamic nature of topics, and responsible opinion formation among people searching the web for information on emerging topics. To realize this vision, we propose three overarching research questions, aimed at understanding the status quo, determining requirements of systems aligned with our vision, and building these systems. For each of the three questions, we highlight relevant literature, including pointers on how they could be addressed. Lastly, we discuss the challenges that will potentially arise in pursuing the proposed vision.

Abstract
This work proposes a bid shading strategy for first-price auctions as a measure-valued optimization problem. We consider a standard parametric form for bid shading and formulate the problem as convex optimization over the joint distribution of shading parameters. After each auction, the shading parameter distribution is adapted via a regularized Wasserstein-proximal update with a data-driven energy functional. This energy functional is conditional on the context, i.e., on publisher/user attributes such as domain, ad slot type, device, or location. The proposed algorithm encourages the bid distribution to place more weight on values with higher expected surplus, i.e., where the win probability and the value gap are both large. We show that the resulting measure-valued convex optimization problem admits a closed form solution. A numerical example illustrates the proposed method.

Abstract
Cyber defense operations increasingly require long-term strategic planning under uncertainty and resource constraints. We propose a new use of combinatorial auctions for allocating defensive action bundles in a realistic cyber environment, using host-specific valuations derived from reinforcement learning (RL) Q-values. These Q-values encode long-term expected utility, allowing upstream planning. We train CAFormer, a differentiable Transformer-based auction mechanism, to produce allocations that are approximately incentive-compatible under misreporting. Rather than benchmarking against existing agents, we explore the qualitative and strategic properties of the learned mechanisms. Compared to oracle and heuristic allocations, our method achieves competitive revenue while offering robustness to misreporting. In addition, we find that allocation patterns correlate with adversarial and defensive activity, suggesting implicit alignment with operational priorities. Our results demonstrate the viability of auction-based planning in cyber defense and highlight the interpretability benefits of RL-derived value structures.

Abstract
How does targeted advertising influence electoral outcomes? This paper presents a one-dimensional spatial model of voting in which a privately informed challenger persuades voters to support him over the status quo. I show that targeted advertising enables the challenger to persuade voters with opposing preferences and swing elections decided by such voters; under simple majority, the challenger can defeat the status quo even when it is located at the median voter's bliss point. Ex-ante commitment power is unnecessary -- the challenger succeeds by strategically revealing different pieces of verifiable information to different voters. Publicizing all political ads would mitigate the negative effects of targeted advertising and help voters collectively make the right choice.

Abstract
We introduce and analyze a discrete soft-decision channel called the linear reliability channel (LRC) in which the soft information is the rank ordering of the received symbol reliabilities. We prove that the LRC is an appropriate approximation to a general class of discrete modulation, continuous noise channels when the noise variance is high. The central feature of the LRC is that its combinatorial nature allows for an extensive mathematical analysis of the channel and its corresponding hard- and soft-decision maximum likelihood (ML) decoders. In particular, we establish explicit error exponents for ML decoding in the LRC when using random codes under both hard- and soft-decision decoding. This analysis allows for a direct, quantitative evaluation of the relative advantage of soft-decision decoding. The discrete geometry of the LRC is distinct from that of the BSC, which is characterized by the Hamming weight, offering a new perspective on code construction for soft-decision settings.

AI Insights

• Soft‑decision critical rate Λ′Z(1)=h(t(β)), where t(β) maximizes αh(t)+J(r_t,β;β)−t ln r_t,β over t∈[0,1].
• Envelope theorem gives Λ′Z(α)=h(t(α,β)), exposing a simple dependence on the maximizer.
• t(β)=1/(β W₀[β e^β/(e^β−1)]) links the LRC to the Lambert W function.
• Jensen’s inequality on concave h(q(t)) proves Λ′N(1)<Λ′Z(1).
• Positivity of f(z)=z²−(ln(1+z/2))/(ln z)−3z²−2z−¼ ensures t(β)
• Soft‑decision decoding thus outperforms hard decoding in high‑noise regimes.
• These insights supply a toolkit for random‑code design in the LRC’s combinatorial geometry.

Abstract
Bayesian Federated Learning (BFL) combines uncertainty modeling with decentralized training, enabling the development of personalized and reliable models under data heterogeneity and privacy constraints. Existing approaches typically rely on Markov Chain Monte Carlo (MCMC) sampling or variational inference, often incorporating personalization mechanisms to better adapt to local data distributions. In this work, we propose an information-geometric projection framework for personalization in parametric BFL. By projecting the global model onto a neighborhood of the user's local model, our method enables a tunable trade-off between global generalization and local specialization. Under mild assumptions, we show that this projection step is equivalent to computing a barycenter on the statistical manifold, allowing us to derive closed-form solutions and achieve cost-free personalization. We apply the proposed approach to a variational learning setup using the Improved Variational Online Newton (IVON) optimizer and extend its application to general aggregation schemes in BFL. Empirical evaluations under heterogeneous data distributions confirm that our method effectively balances global and local performance with minimal computational overhead.

Abstract
In human-computer interaction applications like hand gesture recognition, supervised learning models are often trained on a large population of users to achieve high task accuracy. However, due to individual variability in sensor signals and user behavior, static models may not provide optimal performance for all users. Personalizing pretrained models via calibration--collecting labeled data from each user--can improve performance but introduces user friction and struggles with limited data. To overcome these issues, we propose a calibrationless longitudinal personalization method: a contextual multi-arm bandit (MAB) algorithm combined with a pretrained neural network for gesture recognition. This reinforcement-learning-style approach enables personalization using binary reward signals, either user-provided or inferred by the system. We validated this method in a user study. Participants wore a surface electromyography (sEMG) device and played multiple rounds of a 2-D navigation game using six hand gestures. In the session, they completed a baseline round and then a round with our algorithm; in the second session, they played another round with our algorithm. Our approach led to a significant reduction in users' average false negative rate by 0.113 from the initial to the final round, with further decreases between sessions. Average precision also trended upward (by 0.139) from the start to end of a round, continuing in the next session. Notably, some users who could not complete the game with the baseline model succeeded with our contextual MAB model. In summary, our

AI Insights

• The algorithm casts each gesture as a bandit arm, updating its policy online with binary success/failure rewards.
• It fuses implicit system confidence and explicit user clicks, enabling calibration‑free personalization.
• Across two sessions, false negatives fell (p = 0.002) while precision rose by 0.139 per round.
• Users who failed the baseline game reached 100 % success after one learning round, rescuing edge‑case performers.
• The sEMG‑based 2‑D navigation task shows bandits adapt to highly variable muscle signals across individuals.
• A limitation is the linear reward‑feature assumption, which may miss complex gesture dynamics.

Abstract
Large Language Models (LLMs) have shifted in just a few years from novelty to ubiquity, raising fundamental questions for data science education. Tasks once used to teach coding, writing, and problem-solving can now be completed by LLMs, forcing educators to reconsider both pedagogy and assessment. To understand how instructors are adapting, we conducted semi-structured interviews with 42 instructors from 33 institutions in 10 countries in June and July 2025. Our qualitative analysis reveals a pragmatic mix of optimism and concern. Many respondents view LLMs as inevitable classroom tools -- comparable to calculators or Wikipedia -- while others worry about de-skilling, misplaced confidence, and uneven integration across institutions. Around 58 per cent have already introduced demonstrations, guided activities, or make extensive use of LLMs in their courses, though most expect change to remain slow and uneven. That said, 31 per cent have not used LLMs to teach students and do not plan to. We highlight some instructional innovations, including AI-aware assessments, reflective use of LLMs as tutors, and course-specific chatbots. By sharing these perspectives, we aim to help data science educators adapt collectively to ensure curricula keep pace with technological change.

Abstract
Memory-to-memory data streaming is essential for modern scientific workflows that require near real-time data analysis, experimental steering, and informed decision-making during experiment execution. It eliminates the latency bottlenecks associated with file-based transfers to parallel storage, enabling rapid data movement between experimental facilities and HPC systems. These tightly coupled experimental-HPC workflows demand low latency, high throughput, and reliable data delivery to support on-the-fly analysis and timely feedback for experimental control. Off-the-shelf messaging frameworks are increasingly considered viable solutions for enabling such direct memory streaming due to their maturity, broad adoption, and ability to abstract core messaging and reliability functionalities from the application layer. However, effectively meeting the workflows' requirements depends on utilizing the framework's capabilities and carefully tuning its configurations. In this paper, we present a study that investigates the messaging parameters, and their configuration choices that impact the streaming requirements of two representative scientific workflows. We specifically characterize throughput trade-offs associated with reliable message transmission for these workflows. Our study is conducted through streaming simulations using synthetic workloads derived from the Deleria and LCLS workflows, employing the RabbitMQ messaging framework within the context of the Data Streaming to HPC infrastructure at OLCF. Our simulations reveal several key observations and practical insights that help users understand which configurations best meet the needs of their streaming workloads.

AI Insights

• The study pioneers AI‑coupled HPC workflows that fuse machine‑learning inference with real‑time data streams, slashing analysis latency by an order of magnitude.
• It proposes a unified, API‑driven research infrastructure that stitches together experimental instruments, data‑caching layers, and HPC back‑ends into a single, secure pipeline.
• The authors demonstrate how deep‑learning pipelines can be embedded directly into the streaming fabric, enabling on‑the‑fly feature extraction for materials science and neutron crystallography.
• A key insight is that secure, role‑based API gateways not only protect sensitive experimental data but also reduce overhead by eliminating redundant authentication hops.
• The paper highlights that achieving seamless transitions from lab to production HPC requires coordinated tooling, not just high‑bandwidth links, and recommends a modular plug‑in architecture.
• The authors caution that the proposed framework demands significant infrastructure investment and skilled personnel, underscoring the need for community‑wide tooling standards.
• Finally, the study calls for collaborative R&D to refine the AI‑coupled workflow model, suggesting that shared benchmarks could accelerate adoption across scientific domains.

Abstract
We examine the relative timeliness with which write-downs of long-lived assets incorporate adverse macroeconomic and industry outcomes versus adverse firm-specific outcomes. We posit that users of financial reports are more likely to attribute adverse firm-specific outcomes to suboptimal managerial actions, which provide managers with more incentive to delay write downs. We provide evidence that, controlling for other incentives to manage earnings, firms record write-downs in the current year that are driven by adverse macroeconomic and industry outcomes during both the current year and the next year, but they record write-downs driven by adverse firm-specific outcomes only in the current year.

Abstract
The rapid growth of digital and AI-generated images has amplified the need for secure and verifiable methods of image attribution. While digital watermarking offers more robust protection than metadata-based approaches--which can be easily stripped--current watermarking techniques remain vulnerable to forgery, creating risks of misattribution that can damage the reputations of AI model developers and the rights of digital artists. These vulnerabilities arise from two key issues: (1) content-agnostic watermarks, which, once learned or leaked, can be transferred across images to fake attribution, and (2) reliance on detector-based verification, which is unreliable since detectors can be tricked. We present MetaSeal, a novel framework for content-dependent watermarking with cryptographic security guarantees to safeguard image attribution. Our design provides (1) forgery resistance, preventing unauthorized replication and enforcing cryptographic verification; (2) robust, self-contained protection, embedding attribution directly into images while maintaining resilience against benign transformations; and (3) evidence of tampering, making malicious alterations visually detectable. Experiments demonstrate that MetaSeal effectively mitigates forgery attempts and applies to both natural and AI-generated images, establishing a new standard for secure image attribution.