Machine Learning
Evaluating Open-Weight E-Commerce Agents with Environment-Grounded Verification
Nimit Shah and Haitz Sáez de Ocáriz Borde
A shopping conversation has many routes to the same cart, and a task-success rate reduces all of them to one score. We build a deterministic and reproducible e-commerce environment that precommits each trial's customer and trajectory parameters, including the persona, difficulty, target cart, and an item reveal schedule. A simulated consumer attempts to buy a target cart from the environment with assistance from the evaluated model. The environment guides the simulator's actions and records every assistant action alongside the environment state at that point. After the trial, these records allow the evaluator to assess individual parts of the conversation against the retained evidence. For example, the evaluator penalizes a search for failing to surface a target product only when the customer has already mentioned that product. We further use this evidence to apply different penalties to tool calls depending on how the assistant's actions compare with an expected tool-call set. Our environment also interacts with the simulator bidirectionally, reading its output to stop the trial when the simulator determines that the customer has become too frustrated and injecting directives in real time that specify when to explore, defer buying an item, or recall a previous exchange. This interaction creates an open-ended and verifiable simulation. Across eight open-weight
agents from 20B to 35B parameters, with 160 trials per
agent and 44 metrics, the resulting capability profiles distinguish under-action, over-purchase, unsupported product attributes, and poor search, all of which terminal success obscures.
Read the full abstract ▾
Machine Learning
Distilling Foundation Models for Agentic What-If Reasoning:Cost, Latency, and Governance in a Hybrid LLM+SLM Architecture
Sourish Dey and Aditya Kumar
Tabular foundation models deliver strong zero-training predictive performance via in-context learning, but their high inference latency makes them impractical as hot-path decision backends in interactive agentic loops. We distill a TabPFN teacher into a compact feed-forward student across a business-decision simulation on UCI Adult and five OpenML
benchmarks: the classification head compresses 53.2M parameters to 8,546 (6,220x); the deployed two-head loan pipeline compresses 111.4M parameters to 17,059 (6,532x). The student retains 95.4-100.5% accuracy and 96.8-100.0% AUC, with the lowest accuracy retention on credit-g at 95.4%; an alpha = 0 hard-label control shows that the teacher's soft targets provide a 2.1-7.0 AUC point gain.
Read the full abstract ▾
Sound
MUUNRiver-Bench: Diagnosing Relation-Dependent Music Retrieval with Multimodal Instructions
Zhancheng Guo, Congren Dai, Shangda Wu et al. (+4 more)
Music
retrieval is relation-dependent: given a reference track, a listener may seek its style with a new theme, a cover, or a comparable voice, and these intents demand contradictory rankings. We present MUUNRiver-Bench, a diagnostic
benchmark whose reference-audio queries use natural-
language instructions to define relevance. A pipeline combining expert genre priors, LLM-generated
prompts and lyrics, synthesis, and expert review yields 3,440 tracks spanning 13 genres and 116 sub-genres, and seven tasks: similar-music, style-preserving lyric-rewriting, lyric-preserving style-rewriting, cover, vocal-timbre, isolated-vocal, and segment
retrieval. Across six models in eight configurations, task-wise rank reversals reveal complementary biases: acoustic encoders favour local identity, whereas text-aligned encoders favour semantic relations. Frozen encoders diagnose default similarity preferences; instruction-aware and audio-text fusion systems provide exploratory tests of textual conditioning, with neither simple fusion scheme consistently improving its backbone
Read the full abstract ▾
All authors Zhancheng Guo, Congren Dai, Shangda Wu, Jianhuai Hu, Danni Zhao, Xiaobing Li, Maosong Sun
Robotics
Structure-Preserving Quantum Circuit Architectures for Robot Kinematics
Andrea Morghen, Pierluigi Arpenti, Roberto Schiattarella et al. (+2 more)
Structured spatial data require
quantum encodings that preserve
geometric relations, expose measurable observables, and remain implementable on finite-depth hardware. This work introduces a
quantum representation and
circuit architecture for rigid-body transformations and specializes it to Denavit–Hartenberg kinematics of serial open-chain manipulators. Each translational contribution is factorized into a classical metric magnitude and a signed unit direction encoded by a single-
qubit Bloch vector, while parameterized rotations reproduce the ordered propagation of frame directions. A selector register prepares probabilities proportional to the contribution magnitudes, and the reduced state of a designated readout
qubit encodes their normalized weighted sum. The retained classical scale then reconstructs the metric end-effector position. Two additional readout
qubits encode terminal-frame axes, providing a compact and geometrically interpretable pose interface. At the ideal expectation-value level, measured Pauli observables reproduce the corresponding classical kinematic quantities. Alternative
circuit architectures realize the same representation with different tradeoffs in
qubit count,
circuit depth, controlled operations, and measurement requirements. Validation on a serial manipulator yields numerically negligible position and orientation reconstruction errors under ideal simulation. Finite-shot simulations, noisy executions, transpilation analysis, and a hardware demonstration further characterize statistical error, noise sensitivity, and implementation overhead without asserting computational advantage.
Read the full abstract ▾
All authors Andrea Morghen, Pierluigi Arpenti, Roberto Schiattarella, Giovanni Acampora, Bruno Siciliano
Social and Information Networks
netseg: a Python Package for Measuring Structural Polarization and Segregation in Social Networks
Onur Tuncay Bal and Michał Bojanowski
The study of structural polarization and segregation in social networks is an established line of research, and the quantification of both phenomena proceeds through a set of widely cited network indices. The code implementing those indices, however, is seldom released and almost never tested. We present netseg, a comprehensively documented Python package implementing these indices, most of them generalized to more than two groups and to directed as well as undirected input. It ports the R package of the same name and adds measures the R version lacks, among them Random Walk Controversy, Boundary Connectivity, Dipole Moment, and Moran's I. The package operates on igraph objects and performs the underlying graph operations (e.g., neighborhood queries and random-walk simulation) through igraph's Python interface, so that they execute in compiled code rather than in interpreted Python. For several of the indices this yields runtimes orders of magnitude below those of the available open-source implementations, which the documentation reports in
benchmarks. Most of these indices are defined as a divergence from a null model, and published implementations fix that null model to a uniform random graph of matching density. netseg accepts an ensemble of graphs as a sample from an arbitrary null model, and distinguishes indices that already incorporate a baseline from those that do not, adjusting the comparison accordingly to avoid double subtraction. The documentation provides, for each index, a worked empirical example, its behaviour at the degenerate cases where it is undefined, a
benchmark, and the procedure for substituting a custom null model. We report the behaviour of every index over a parameter sweep of a generative opinion model, and apply them to a county-level railroad network built from nineteenth-century operator records joined to full-count census data.
Read the full abstract ▾
Computers and Society · Cross-listed from physics.soc-ph
How to build a campfire? Participatory modelling with justice
Nynke van Uffelen, Sander ten Caat, Aarthi Sundaram et al. (+4 more)
Energy transition decision-making is pervaded by deep uncertainties. Computational models are helpful in addressing such uncertainties, as they can give insight into techno-economic complexity. However, models alone are insufficient, as part of the uncertainties involve justice dilemmas.
Read the full abstract ▾
All authors Nynke van Uffelen, Sander ten Caat, Aarthi Sundaram, Annemiek de Looze, Marion Collewet, Eefje Cuppen, Igor Nikolic
Hardware Architecture
Is INT8 Portable? A Cross-Platform Measurement Study of Quantized Inference on Embedded and Automotive Accelerators
Yuyeong Shin
Eight-bit integer (INT8) post-training quantization is the default recipe for edge deployment, under a widely held assumption: INT8 makes inference faster at a small, predictable accuracy cost, and a model quantized once can be carried to any target. We test that assumption with a controlled measurement study across seven hardware classes – ARM and x86 CPUs, a discrete GPU, an NVIDIA Jetson AGX Orin iGPU and its NVDLA cores, and two vendor NPUs (Qualcomm Hexagon HTP, DEEPX DX-M1) – holding the ONNX artifact and the quantization scales fixed so the integer kernel or ISA is the only free variable. Portability fails on three axes. (1) The sign of the INT8 speedup is set by the CPU's dot-product ISA (ARM dotprod/SDOT, x86 VNNI): cores that have it speed up by up to 2.1x, cores that lack it slow down by 1.7x, for the identical model and runtime. (2) INT8 outputs are not portable, and the rule is an invariance rather than a gradient: FP32 predictions are bit-identical for every pair (1000/1000), while INT8 predictions agree 1000/1000 exactly when two targets share an integer kernel and 958-965/1000 whenever they do not – independent of whether the boundary is CPU<->CPU or CPU<->accelerator, and invisible to top-1 accuracy, which is preserved. (3) Vendor NPUs own quantization: a bring-your-own QDQ graph fails silently on one NPU (external scales ignored, accuracy 0.75 -> 0.005 while it compiles, profiles and runs without error) and loudly on the other (the compiler refuses the graph), so only the vendor's native path yields a correct engine. We further show that edge-NPU latency regimes are set by output/device-to-host transfer size rather than compute, and locate the transition with a fixed-compute sweep. We release the scripts and 32 reports. "Quantize once, deploy anywhere" is unsafe for embedded and automotive deployment, where per-input determinism and redundancy matter.
Read the full abstract ▾
Social and Information Networks
Predicting Social Media Engagement using Machine Learning
Ritwik Singh, Mayukh Majumdar and Subodha Kumar
Social media platforms are popular channels for disseminating information, owing to their large user bases and ease of access. Companies also use social media as an important aspect of the advertising process. By creating high-quality posts, companies can strengthen their engagement metrics and increase their follower count. While a growing body of research has examined social media engagement, fewer studies have jointly examined the visual, textual, and temporal features of image posts, even though these features collectively determine the performance of content on social media. To understand the important drivers of social media engagement, we collect image posts of furniture firms on Facebook and extract visual, temporal, and textual features from them using text and image analytics methods. We evaluate several machine learning models–including Random Forest, Light Gradient Boosting Machine (LightGBM), and eXtreme Gradient Boosting (XGBoost)–to assess the drivers and the prediction power of social media engagement using the features from our data. Our primary model, LightGBM, explains 81.8% of the variance in the engagement score, 2.3 times that of our baseline model. We observed that brand-relative deviation features, which capture how far a post departs from its brand's norms, are the strongest predictors, accounting for 51.3% of total importance. Our research quantifies the extent to which these features are associated with interactions and provides recommendations that organizations may consider.
Read the full abstract ▾
Artificial Intelligence
Corrupt Plans, Clean Traces: Evading Chain-of-Thought Monitoring with Plan Injection
Keertana Chidambaram, Andrew Ilyas and Vasilis Syrgkanis
Chain-of-thought (CoT) monitoring is a safety strategy where the
reasoning of a large
language model "actor" is inspected by a "monitor" (often another
language model) for signs of unsafe planning, deception, or misalignment. We find that planting harmful but benign-sounding
reasoning in the actor's context can steer it to perform
adversarial actions while evading monitors, an attack we term "plan injection". We initially discover this attack in the multiple-choice question-answering monitorability setting proposed by Lanham et al. (2023), using the investigator-
agent elicitation framework of Li et al. (2025). We generalize the attack and show that the discovered behavior scales to harder tasks (achieving 25-33% monitor evasion rates across different monitorability
benchmarks) and larger models such as DeepSeek-R1. Across the settings we study, actor models not only follow injected plans but also paraphrase them as their own
reasoning, without explicit attribution to the injections. Finally, we find cases where extra monitor resources cause harm - giving the monitor access to the injected plan drops detection by as much as 50% in the Bio-Math task and in a case study on monitor
reasoning budget, we find transcripts where additional thinking
tokens are spent rationalizing the injected plan rather than flagging it.
Read the full abstract ▾
Robotics
ResSafe: Learning Safety Filtering with Residual Reinforcement Learning for Humanoids
Gechen Qu, Tong Zhang, Bike Zhang et al. (+4 more)
Safe control of
humanoid robots remains challenging due to their high-dimensional dynamics, contact-rich interactions, and sensitivity to disturbances. Although reinforcement learning has enabled effective locomotion and motion tracking, learned policies can still generate unsafe actions that lead to instability or falls. In this work, we propose residual reinforcement learning as an implicit safety-filtering mechanism for safe
humanoid control. Instead of relying on a single nominal policy to simultaneously balance performance, safety, and robustness, we decouple performance and safety. The nominal policy focuses solely on task performance, while a residual policy learns safety corrections. This decoupling leads to a better performance–safety Pareto trade-off and avoids the need for careful tuning of multiple competing reward terms within a single policy training. We show that the residual policy can act as an implicit safety filter.
Read the full abstract ▾
All authors Gechen Qu, Tong Zhang, Bike Zhang, Yen-Jen Wang, Koushil Sreenath, Claire Tomlin, Jason Jangho Choi
Machine Learning
Bellman Policy Optimization
Zhuoqing Song, Haotian Xu, Xikun Zhang et al. (+1 more)
Reinforcement learning with verifiable rewards (RLVR) improves the
reasoning capabilities of large
language models (LLMs). We introduce Bellman Policy Optimization (BPO), a critic-free method derived from Policy Mirror Descent (PMD). For autoregressive generation with terminal rewards, BPO uses the Bellman equations to reformulate PMD as a trajectory-level objective. The reformulation avoids estimating state values at intermediate states. We prove that it has the same unique optimal solution as the original PMD objective. We derive the practical BPO loss by approximating this objective. Its mismatch-correction weight is a smoothed ratio of complementary
token probabilities. Experiments on mathematical
reasoning benchmarks demonstrate the effectiveness of BPO.
Read the full abstract ▾
All authors Zhuoqing Song, Haotian Xu, Xikun Zhang, Lidong Bing
Artificial Intelligence
Stellar Colosseum: A Many-Agent Harness for Long-Horizon Research in Mathematics and Theoretical Computer Science
Honghao Lin, David P. Woodruff, Yuan Deng et al. (+3 more)
Language models can produce plausible short proofs, but may still be unreliable on long-horizon research problems, where progress depends on a sequence of uncertain and interdependent decisions. We introduce
Stellar Colosseum, a model-agnostic harness for allocating inference across research in mathematics and theoretical computer science. Colosseum explores alternative strategies before proof construction, uses a readiness gate to decide when a route is mature enough to decompose, represents the proof plan as interdependent section-level subproblems, and routes verifier findings back to the affected part of the argument. Across these stages, it generates candidates in parallel, attacks them with targeted falsification, and combines candidates and their critiques into a single research artifact through overlapping random-sample tree aggregation. The Colosseum workflow has been integrated into Google Antigravity's Teamwork framework as the Long Proof pattern. We demonstrate the capabilities of Colosseum through open-ended research and evaluations on
theorem-proving and competitive programming
benchmarks. Using Colosseum with Gemini 3.1 Pro, we obtain several new results that address open problems arising from papers published at top venues such as FOCS and JMLR. On TCS-Bench, a
benchmark of research-level
theorem-proving tasks drawn from papers published at FOCS, STOC, and SODA, Colosseum achieves 71.0% accuracy using Gemini 3.1 Pro and Gemini 3.7 Flash. In a separate Codeforces evaluation using Gemini 3.1 Pro, the proof-oriented pipeline with execution feedback solves 218 of 222 problems.
Read the full abstract ▾
All authors Honghao Lin, David P. Woodruff, Yuan Deng, Jieming Mao, Song Zuo, Vahab Mirrokni
Machine Learning
The Router Within: Eliciting Native Skill Routing from a Frozen LLM
Ruishuo Chen, Xun Wang, Yu Chen et al. (+2 more)
Skills extend an LLM
agent beyond its parametric knowledge, and the gain they promise rests on picking the right one. Deployed harnesses route by preloading every skill's metadata into the context, which disperses the
agent's
attention and caps the library size.
Retrieval pipelines move the selection out of the context, but also out of the
agent's capability. We show that the frozen
agent LLM already carries the routing signal in its own forward passes, and that two linear maps suffice to read it out with no skill text in the context. Our Gavel (Glance And Verdict from a frozen LLM) reads it in two steps. A glance scores the full library by matching the task's mid-layer states against a compact bank that one forward pass builds for each skill at installation, with the two maps as the only trained parameters. A verdict then resumes each shortlisted skill's forward pass, reads the model's own likelihood and yes/no judgment, and fuses both with the glance as a product of experts. Trained once, Gavel transfers zero-shot to three public
benchmarks and SkillTraj, our new
benchmark of 372 simulated
agent trajectories. On Qwen3-32B it outperforms progressive disclosure and retrieve-and-rerank pipelines that add 1.2B to 16B external parameters, by up to 13.4 points on written tasks and up to 21.9 when the need for a skill arises mid-rollout. Routing accuracy improves as the backbone does, and in a bash-
agent harness Gavel lets the 32B trigger the right skill on Skill-Use more often than models of up to 1.6T parameters in Codex.
Read the full abstract ▾
All authors Ruishuo Chen, Xun Wang, Yu Chen, Zhuoran Li, Longbo Huang
Machine Learning
A Chosen Future Can Still Be Rewritten: Causal Writability in Video Models
Xingyun Wang, Haomin Zheng, Man Yuan et al. (+2 more)
When a
video model generates physically incorrect motion, did it fail to learn the correct motion, or did it learn it but fail to use it? We show the latter: the correct motion remains available inside the model and can still be made to control the generated
video. We train on
videos where red masses oscillate slowly and blue masses oscillate quickly, then test a red mass with fast observed motion. Even when the model generates slow motion in this conflicting case, a low-dimensional edit predicted from simple physical variables restores the correct fast motion. We call this ability causal writability. At fixed strength, we find a sharp depth boundary: the same edit changes the
video before the boundary but not after it. This closure marks commitment for that write. The motion signal nevertheless remains, and a stronger downstream write can restore physical motion, while excessive gain overshoots. Early causal writability predicts which errors training later corrects: those errors are writable at more network depths than errors that persist. We reproduce both causal writability and its sharp closure in a pretrained 1.3B
video model, supporting generality across model scale and training regime.
Read the full abstract ▾
All authors Xingyun Wang, Haomin Zheng, Man Yuan, Leqian Yang, Ziming Liu
Data Structures and Algorithms
The $k$-server conjecture is true
Christian Coester, Elias Koutsoupias and Marek Zbysiński
The $k$-server
conjecture states that a deterministic online
algorithm can achieve competitive ratio $k$ on every metric space. We prove the
conjecture. Specifically, we show that the work function
algorithm satisfies it. Our proof uses a natural algebraic representation of the work function as a
matrix, which encodes all feasible paths to reach a configuration. In this representation, the minimum and addition operations arising in the definition of optimal costs correspond to addition and multiplication of formal expressions, and each work function value corresponds to the determinant of $k$ columns of the
matrix. A request arrival updates the representation via a change of basis and row replacement. The amortized analysis is based on a potential function defined in terms of a larger
matrix whose coordinates are pairs of coordinates of the original
matrix representation.
Read the full abstract ▾
Human-Computer Interaction
The CAST-framework: Measure and model social media use as a multi-level phenomenon through real-world applications
David Grüning, Jasper Doeninghaus, Zina Efchary et al. (+8 more)
Designing social media experiences that support well-being requires understanding when, how, and for whom use matters. Screen-time totals omit content and context, and connecting these with behavior and experience requires coordinating measurements across timescales. We introduce the CAST framework to connect measurement choices with person-specific models of exposure, behavior, physiology, and experience. Its dimensions specify where observations occur, how they are obtained, what they measure, and at what temporal resolution. Responses to interventions, such as whether to proceed after an app-opening pause, enter as behavioral measurements. We propose four synchronized measurement modules linking mobile and wearable data with self-reports and intervention responses. A synthetic demonstration with 120 simulated participants over 28 days illustrates how daily aggregation can obscure opposing effects of different activities under specified generating assumptions. The framework guides selection of measures and outcomes for evaluating social media interfaces and interventions.
Read the full abstract ▾
All authors David Grüning, Jasper Doeninghaus, Zina Efchary, Yui Kondo, Kevin Dunnell, Lennart Fischer, Isabella Zimmermann, Linnea Körte, Leo Mehlig, Frederik Riedel, Paul Schmiedmayer
Information Theory
A Hardy-Space Proof of the Filter-Only Gaussian Feedback-Capacity Formula
Jun Su, Yang Xu and Guangyue Han
The feedback capacity of power-constrained channels with additive stationary Gaussian noise was formulated by Kim in [Kim, 2010] as an infinite-dimensional optimization over the spectrum of an independent stationary Gaussian component and a strictly causal feedback filter. Kim further asserted that the independent stationary component can be omitted. However, Derpich and Østergaard [Derpich and Østergaard, 2022] identified a gap in the original proof, leaving the original derivation of the filter-only capacity formula incomplete. In this paper, we establish the filter-only formula at the level of capacity suprema for noise spectra satisfying the Paley-Wiener condition. In particular, we show that
polynomial feedback filters suffice in the capacity supremum, without a positive lower bound on the noise spectrum or a rationality assumption. Under a positive essential lower bound, a complementary approximation preserves the output spectrum and rate exactly, while its input power converges to that of the original pair. An interleaved AR(1) example connects these results with the SK(2) coding scheme. The argument does not assert attainment by a single filter.
Read the full abstract ▾
Robotics
MessyMem: Learning-from-Doing Memory for Mobile Manipulation
Anuva Banwasi, William Muckelroy III, Priya Sundaresan et al. (+3 more)
Mobile manipulators deployed across many rooms and visits should improve with experience: after discovering that a cabinet is locked or finding an object in a drawer, the
robot should reuse that knowledge rather than start each task from scratch. Yet today's
robots often treat each task as new: compact scene representations omit interaction-derived knowledge, raw
video histories are difficult to query, and VLM planners reason at inference time without persistently updating what the
robot knows. We present MessyMem, a persistent
memory system that enables mobile manipulators to learn from experience and reuse that knowledge across future tasks. It maintains a spatially grounded 3D scene graph of objects and locations, augments it with properties and outcomes learned through interaction, and links visual observations for fine-grained recall. We evaluate MessyMem in simulation and on a real mobile manipulator. In a continuous 25-task simulation spanning over 3 hours, MessyMem achieves 80.0% task progress, outperforming the strongest ablation by 14.8 percentage points and the strongest external baseline by 28.9 points, while retrieving task-relevant evidence from thousands of stored keyframes and over an hour into the past.
Read the full abstract ▾
All authors Anuva Banwasi, William Muckelroy III, Priya Sundaresan, Linfeng Zhao, Jeannette Bohg, Cherie Ho
Computation and Language
Disentangling Representation Evolution in Transformers through Directional Decomposition
Shwai He, Haichao Zhang and Shen Yan
Transformer representations evolve through learned additive transformations that either preserve their current direction or redirect it. We study this
evolution as a functional geometry, decomposing learned updates into parallel and perpendicular components. Across pretrained models, we find substantial parallel components beyond the residual identity path. We then apply the decomposition in two spaces: to
attention and MLP updates relative to the hidden state, and to
attention value aggregation relative to the current
token's value. Targeted edits reveal a strongly space-dependent asymmetry: exclude-self value-space parallel manipulation is markedly more robust than residual-space and perpendicular counterparts, preserving the direct self message while scaling only the non-self aggregate. The same decomposition gives a component-resolved description of compression-induced update error: perpendicular error separates compression methods more clearly than parallel error. Extensive experiments further demonstrate that full-aggregate parallel suppression during from-scratch pretraining lowers validation-loss trajectories and improves downstream averages, with the value-space variant strongest. Together, these results connect representation geometry to editing robustness, compression diagnosis, and training-time intervention. Code is available in the \href{https://github.com/Shwai-He/
Transformer-Geometry}{project repository}.
Read the full abstract ▾
Computation and Language
Discovery Foundation Models: Toward Open-Ended Discovery Intelligence
Ling Yang, Zhenfei Yin and Yingcheng Wu
Foundation models have progressed from learning and
reasoning over existing knowledge, to increasingly learning through action, tool use, and outcome feedback. We argue that the next frontier is a further transition: from solving and acting within problems specified by humans to participating in the process by which new problems, representations, explanations, and knowledge are created. We refer to this capability as Discovery Intelligence. We formulate Discovery Foundation Models (DFMs) as general-purpose model systems for open-ended discovery. A DFM operates over a revisable research state and supports seven coupled capabilities spanning problem discovery, formulation, representation construction, hypothesis formation, intervention, evidence-grounded revision, and continual discovery improvement. We instantiate this framework with Zetema, which couples explicit research-state dynamics, verification and experimental gating, external grounding, and cross-task Discovery Skill
evolution. We further ground the framework with GALILEO, a real therapeutic-discovery system in which Dry-Lab
reasoning, robotic and hands-on Wet-Lab experimentation, external biological evidence, and iterative hypothesis and design revision form a closed physical discovery loop. We then formulate a unified approach to capability formation and process-centered evaluation, enabling discovery behavior to be trained, improved, and measured beyond final-answer performance. Together, these components establish discovery as a learnable, executable, and evaluable capability of foundation-model systems. We view this shift as a broader progression in intelligence scaling: from learning over existing knowledge, to learning from action outcomes, and ultimately to participating in the construction, testing, and revision of the structures through which new knowledge is discovered. Code: https://github.com/Gen-Verse/DFM-Plans
Read the full abstract ▾
Computation and Language
Mind2Dialogue: Training Human-Aware Language Models by Simulating User Mental States
Zixuan Wang, Yufan Zhou, Jinzhou Tang et al. (+16 more)
As
language models become more capable, long-term collaboration in learning,
reasoning, and decision-making calls for a deeper understanding of the people they serve. Yet training such human-aware
language models faces a fundamental supervision gap because current
datasets for LLM assistant training contain few if any well-informed responses explicitly grounded in users' unspoken beliefs and goals. Scaling such supervision is inherently constrained, as users' underlying states are not directly observable. We thus propose the Mind2Dialogue framework to mitigate this gap by simulating users' mental states and turning them into privileged supervision for human-aware training. Specifically, we first propose a psychology-guided simulator that preserves personal characteristics while updating mental states through interaction to generate coherent conversations. The key idea is to enforce a shared evolving mental state that drives user behavior and guides an Oracle assistant's responses. Our privileged
distillation then trains models on the Oracle's well-informed responses to assist users without direct access to their mental states at deployment. Moreover, we propose to evaluate human-aware learning by combining personalization and theory of mind, examining how models understand people and act on that understanding. Training on the full Mind2Dialogue corpus improves every reported personalization metric over the corresponding Qwen, Llama, and OLMo instruction-tuned baselines, including gains of 26.6 to 40.9 percentage points in preference-following generation. The gains extend to belief and action
reasoning on Qwen and Llama, beyond personalized assistance. Looking forward, Mind2Dialogue makes user simulation a foundation for genuine AI collaborators that understand beliefs and intentions behind people's words and support their long-term goals across education, work, and everyday life.
Read the full abstract ▾
All authors Zixuan Wang, Yufan Zhou, Jinzhou Tang, Xinle Yu, Chengjun Wu, Lyumanshan Ye, Zhaoxiang Feng, Letian Peng, Adyasha Patra, Fan Bai, Enze Ma, Zhengding Hu, Jianyang Gu, Zhao Wang, Yufei Ding, Jingbo Shang, Tianmin Shu, Zhiting Hu, Zhen Wang
Computation and Language
Verifiable by Construction: Claim-Level Evaluation of Verbatim Citation in Clinical Question Answering
Jiashuo Zhang, Yuling Chen, Yvonne Commodore-Mensah et al. (+1 more)
Large
language models (LLMs) have been widely adopted for clinical question answering (QA). Current systems can attach citations to their answers, but these often point to broad texts, leaving time-pressed clinicians unable to verify them efficiently. An alternative is to ensure that responses are verifiable by construction: providing fine-grained verbatim quotes from reference material that substantiate claims, so users can verify an answer without opening other documents. In this paper, we evaluate the ability of current models to perform this task end-to-end: from providing citations for every factual claim, to producing verbatim quotes, to ensuring that those quotes fully substantiate the claims. To do so, we build a standardized harness over four clinical practice guidelines and evaluate twelve LLMs on 222 synthetic clinical questions, measuring each of these stages separately. We find that most models can attach verbatim quotes to over 90% of their claims from prompting alone, apart from some lightweight models such as claude-haiku-4.5. Yet these quotes often fail to substantiate every detail of the claims they accompany. For instance, claude-opus-5 produces verbatim quotes for 98.0% of its claims, but fully substantiates only 37.1%. Our work provides insights into the current capability gap of LLMs in building verifiable clinical QA systems, along with artifacts for future research.
Read the full abstract ▾
All authors Jiashuo Zhang, Yuling Chen, Yvonne Commodore-Mensah, Michael Oberst
Cryptography and Security
Adversarial Testing of Automated Program Repair Agents for Security Vulnerabilities
Fares Trad, Simin Chen, Hung Viet Pham et al. (+2 more)
Software
agents with Large
Language Models (LLMs) are designed for Automated Program Repair (APR) tasks, raising the possibility that, in the near future, APR
agents will fix bugs automatically without much human intervention. Can we trust an APR
agent to produce both functionally correct and secure code in such situations? What if attackers target production APR
agents with
adversarial issues that seem benign but may influence the
agents to produce correct but insecure code? In this paper, we took a first step towards answering these questions by conducting an empirical study. First, we created SWEADV, a
benchmark of 750
adversarial issue descriptions constructed from 150 repair tasks in SWE-bench Verified. For each repair task, we created five
adversarial issue descriptions, one for each attack type: command execution, deserialization, path traversal, denial of service, and weak hashing. Second, we evaluated mini_swe APR
agents from three LLM backends on SWEADV: GPT-5-Mini, MiniMax-M2.5, and DeepSeek-R. We found that on average,
adversarial issue descriptions can induce malicious behaviors with successful repair in 51.7% of cases. Third, we investigated whether typical detection mechanisms are sufficient to prevent such malicious patches from being accepted. Pre-repair detection with LLM-as-judge on the
adversarial issue descriptions resulted in an average detection accuracy of only 62.3%. Post-repair detection on
adversarial APR patches using static analysis tools and LLM-as-judge achieved average detection accuracies of only 39.4% and 55.4%, respectively. We conclude that autonomous APR
agents cannot be trusted yet in production deployment, given their susceptibility to
adversarial attacks.
Read the full abstract ▾
All authors Fares Trad, Simin Chen, Hung Viet Pham, Gias Uddin, Baishakhi Ray
Networking
Understanding the oversubscription behaviour of DragonFly+ networks
Vlad-Adrian Ulmeanu, Costin Raiciu and Iulian-Ilie Drăcea
The Max-Host Dragonfly+
topology's original paper proves that there is a 2:1 worst-case oversubscription ratio in expectation for the permutation traffic pattern. We show that the proof only covers a subset of permutation patterns, specifically those in which all host pairs are in different groups, and for any receiver group there are at least two sending hosts in distinct groups. The proof contains two mistakes that cancel out to produce the correct result. Furthermore, the proof implicitly uses an important observation without a backing argument: traffic leaving an indirect group, now forced to follow min-cost paths to receiver groups, can still be split across almost all outgoing global links, in contrast with splitting traffic from a sender group that can naturally select any outgoing group. We prove that the 2:1 ratio holds in expectation for a larger set of patterns, including any permutation. We only need to know that any host sends and receives at most line rate traffic. This constraint can be altered to obtain approximations for expected FCT bounds on any pattern. We also attack the oversubscription problem without the expectation assumption, and find bounds that hold with high probability on permutations for small switch radixes. While in expectation the global layer wiring doesn't affect the oversubscription rate, we experimentally find that it matters in the general case. We find
topologies that obtain visible speedups against the default global wiring from the original paper, under $7\%$ for radixes at most $8$.
Read the full abstract ▾
Logic in Computer Science
Continuity-First Lexicographic Optimization for Home-Care Resource Allocation
Tuyen Van Kieu, Khanh Ngoc Do and Khanh Van To
Home-care allocation must balance continuity of care, caregiver overtime, and caregiver-service compatibility. The published formulation of the Home-Care Optimal Resource Allocation Problem (HCORAP) uses a weighted policy (Weighted) to combine these outcomes, allowing compatibility gains to offset poorer continuity or additional overtime. We introduce LEX-COS, a lexicographic objective for HCORAP that first minimizes continuity violations, then overtime, and finally maximizes compatibility. We give an exact MaxSAT implementation and construct HCORAP-LC, a reproducible 48-instance synthetic evaluation suite with limited caregiver capacity. Compared with Weighted, LEX-COS reduces both continuity violations and overtime in 42 of 48 instances, without increasing either measure in any instance, while yielding lower compatibility. In 36 instances, LEX-COS achieves better continuity and lower overtime than every Weighted-optimal allocation. On the separate 48-instance Original
benchmark, Gurobi and CPLEX, two leading commercial MIP solvers, resolve every instance under both policies and are substantially faster than EvalMaxSAT. EvalMaxSAT also resolves the full
benchmark under LEX-COS. Within EvalMaxSAT, replacing sorting networks with Totalizer improves runtime under Weighted, with no clear gain under LEX-COS. These results characterize the effects of giving continuity and overtime explicit priority in home-care allocation.
Read the full abstract ▾
Machine Learning
Privacy-Aligned Personalized Federated Learning with Compact Adaptation and Variable-Length Gaussian Communication
Yilin Xu, Chun Hei Michael Shiu, Chih Wei Ling et al. (+1 more)
Record-level differential
privacy exposes a structural misalignment in personalized
federated learning when client-specific variation is low-dimensional while training repeatedly releases high-dimensional updates. In this paper, we address this misalignment by releasing a private client context once and confining repeated adaptation to a fixed coefficient space. Beyond dimensionality reduction, the factorized generator induces an adaptive optimization geometry that reshapes noisy updates, and controlled ablations show that most of its private-training gain is retained by radial
evolution. To further reduce the communication cost, we realize the Gaussian mechanism for coefficient updates directly through variable-length quantization with finite expected code length, so that the quantization error itself serves as the required
privacy perturbation rather than extra distortion. Across MNIST and CIFAR-10, our design matches or outperforms full-model private adaptation across
privacy budgets and client heterogeneity, while reducing protected uplink by a factor of 2.67 at \(\varepsilon=16\) on CIFAR-10 with comparable future-client accuracy.
Read the full abstract ▾
All authors Yilin Xu, Chun Hei Michael Shiu, Chih Wei Ling, Linqi Song
Human-Computer Interaction
Pulla: A Parsons Problem Tool for Fine-Grained Behavioral Tracing and Instructor-Facing Problem-Solving Analysis
Daniel Prol, Juho Leinonen, Arto Hellas et al. (+2 more)
Existing Parsons problem tools primarily focus on correctness, indicating whether a student solved a problem, but providing limited visibility into the underlying problem-solving process. We address this gap by introducing Pulla, a Parsons problem tool that instruments programming assignments to capture fine-grained interaction data. These behavioral traces allow the system to surface recurring difficulty patterns, giving instructors actionable insights to inform targeted intervention decisions. This paper describes our experience in developing and deploying Pulla. We deployed the tool in two university courses: an upper-division software design course at the University of Houston (United States) and an introductory programming course at Aalto University (Finland). By analyzing the data collected, we identified common difficulty patterns, including misidentifying exception types, confusing return with the throw/raise mechanism, and incorrect control-flow ordering.
Read the full abstract ▾
All authors Daniel Prol, Juho Leinonen, Arto Hellas, Saleh Alkhamees, Amin Alipour
Robotics
Beyond Single-Axis Testing: Paired Evaluation of Compound Robustness in Vision-Language-Action Policies
Hiroki Sawada and Shunichi Kasahara
Vision-
language-action policies are typically evaluated one perturbation at a time, providing a useful diagnosis of their sensitivity to individual distribution shifts. Real-world deployment, however, may involve several shifts simultaneously, and it remains unclear how these individual robustness measurements compose. We ask whether compound robustness can be inferred from single-axis evaluations. We introduce LIBERO-CTRL, a six-axis
benchmark that pairs each initial state across single-axis conditions and a matched simultaneous condition. This design reveals two opposing outcome changes that aggregate success rates cannot distinguish: emergent failures, where all single-axis rollouts succeed but the simultaneous rollout fails, and compensated successes, where at least one single-axis rollout fails but the simultaneous rollout succeeds. Because one transition decreases compound success while the other increases it, they can cancel, making aggregate compound performance appear consistent with single-axis measurements even when individual outcomes differ substantially. These opposing transitions can largely cancel in aggregate: even when the difference between the two transition rates is not statistically distinguishable from zero, as many as 29.0% of matched initial states still change outcome. Across six policies and three severity levels, such outcome changes reach 34.5% in the most affected condition. The relative prevalence of the two transitions varies across policies and severities, while the transition rates remain similar under independent re-evaluation of
stochastic policies. Compound robustness therefore cannot be characterized from aggregate single-axis success rates alone; matched per-instance evaluation is needed to reveal how joint perturbations alter behavior.
Read the full abstract ▾
Cryptography and Security
Vulnerability Localization Benchmark: Measuring Agentic Security Analysis at Repository Scale
Aman Priyanshu, Supriti Vijay, Kimia Majd et al. (+8 more)
Language-model
agents increasingly operate over complete software repositories, yet cybersecurity evaluations primarily measure whether they can detect, reproduce, or repair vulnerabilities rather than whether they can locate the relevant code. We study vulnerability localization: given a weakness class and an unfamiliar repository, identify the implementation files associated with that weakness. We introduce the Vulnerability Localization
Benchmark (VLoc Bench), comprising 500 real world vulnerabilities from 290 repositories across six package ecosystems and 147 CWE categories. Each task pairs repository snapshots immediately before and after a security fix. On the vulnerable snapshot, an
agent receives only the CWE description and read-only terminal access and must return the affected files; on the patched snapshot, it must determine that the recorded vulnerability is no longer present. We evaluate 27
language models and four static-analysis tools under a common
agent interface. Repository-scale vulnerability localization remains difficult: the strongest system achieves 0.229 File F1, and 38.4% of tasks receive no correct localization from any evaluated model. We further find that stronger localization does not imply reliable behavior after remediation: systems that identify vulnerable files effectively can still report unsupported locations on patched repositories. These results establish vulnerability localization as a distinct repository-scale capability and provide a setting for studying both how security
agents search for vulnerable code and when they should refrain from reporting it.
Read the full abstract ▾
All authors Aman Priyanshu, Supriti Vijay, Kimia Majd, Xuhong He, Fraser Burch, Takahiro Matsumoto, Jianliang He, Baturay Saglam, Arthur Goldblatt, Zhuoran Yang, Amin Karbasi
Computation and Language
HypoEvolve: Genetic Algorithms Enable Multi-Agent LLMs to Discover Scientific Hypotheses
Jieyuan Liu, Mengzhou Hu, Jefferson Chen et al. (+10 more)
Scientific
agents contribute to hypothesis discovery by synthesizing evidence, assessing proposals, and developing new explanations. Recent systems combine scientific
agents with evolutionary search through critique, comparison, and revision. However, how different forms of
agent collaboration affect hypothesis quality remains an open question. Answering this question requires separating the effects of
agents' scientific capabilities from those of their collaboration. A framework must therefore preserve
agents' scientific roles and support rules for combining, revising, and retaining hypotheses. Building on this view, we introduce HypoEvolve, which makes collaboration explicit through successive updates to a hypothesis population. Specifically, we propose a generational genetic
algorithm to coordinate specialized large
language model (LLM)
agents that integrate mechanistic arguments, reconsider assumptions, and assess evidence and testability. Each generation specifies how scientific judgments and new proposals reshape the population, making collaboration effects on hypothesis quality directly testable. Moreover, we design our evaluation around scientifically meaningful hypotheses that explain how a proposed intervention could work. Drug repurposing links these explanations to target-level biological claims assessed against external evidence. Specifically, we adapt DepMap and Open Targets into complementary external measures grounded in experimental, genetic, and clinical evidence. Across 34 cancer types, HypoEvolve achieves the highest scores against six baselines on both measures. DepMap selectivity reaches 0.171, versus 0.115 for the strongest baseline. Gains over single-pass generation also generalize to held-out cancer types. HypoEvolve advances a
vision of autonomous science in which AI research teams achieve a capacity for discovery beyond that of individual models.
Read the full abstract ▾
All authors Jieyuan Liu, Mengzhou Hu, Jefferson Chen, JungHo Kong, Pratibha Jagannatha, Yiming Gao, Dexter Pratt, Hsin-Yuan Lee, Zhiting Hu, Trey Ideker, Wei Wang, Eric P. Xing, Zhen Wang