Papers

Although multivariate time series forecasting has broad application prospects, effectively incorporating exogenous variables remains challenging. Prevailing discrete-time architectures often impose rigid temporal alignment across heterogeneous variables, which may limit their ability to capture continuously evolving external effects. To address this limitation, we introduce CTIA-ODE, a continuous-time forecasting framework centered on Continuous-Time Inter-Variable Attention (CTIA). Specifically, CTIA is embedded within an ordinary differential equation (ODE) solver, enabling the continuously evolving endogenous trajectory to dynamically query a static exogenous context along the integration path without requiring explicit temporal alignment. To improve the stability of the continuous latent dynamics, we further introduce a dual-stabilization strategy: architecturally, a Linear Residual Augmentation (LRA) module provides a macroscopic trend anchor, while an Outlier-Robust Hybrid Loss (ORHL) mitigates the influence of heavy-tailed anomalies during optimization. Extensive experiments across eight real-world datasets demonstrate that CTIA-ODE achieves state-of-the-art forecasting performance across diverse datasets and prediction horizons, while maintaining a favorable accuracy-efficiency trade-off with a compact memory footprint and low inference latency, particularly under long-horizon forecasting settings. Anonymous code is available at https://anonymous.4open.science/r/CTIA-ODE-15A6/.

Cyber Threat Intelligence (CTI) investigators who monitor dark web marketplaces for leaked credentials and breach listings must pass the diverse CAPTCHAs that operators deploy to block automated collection. Existing solvers either need unscalable per-site fine-tuning or rely on commercial Vision-Language Model (VLM) APIs, which surrender operational sovereignty: provider policies, Tor-exit blocking, government regulation, or model updates outside the investigator's control can halt or alter the solver overnight, and per-call costs recur with every session and retry. We present DarKnight, a self-improving agentic framework designed to run on a single workstation GPU inside an investigator's Tor crawler and solve CAPTCHAs with small, locally hosted open-weight VLMs that no provider decision can withdraw, without per-market training or third-party services. For each CAPTCHA, DarKnight retrieves successful and failed past attempts on similar images (dual-signal RAG), composes an image-specific preprocessing pipeline, and learns from the site's pass/fail verdict. In a case study on four real-world dark web marketplaces, agentic tool selection outperformed parameter scaling. A deployable open-weight VLM with DarKnight reached 83.2% on the most heavily defended market, 37.2 percentage points above its base model, outperforming even the frontier Claude Opus~5 (44%). DarKnight recovered 48.2% of baseline failures at a 7.6% degeneration cost, indicating that selective tool use can avoid much of the degradation caused by uniform preprocessing.

This study develops and validates an adaptive parametric method for sizing safety stock when forecast errors are autocorrelated, heteroscedastic, asymmetric, and heavy-tailed. Using four years of weekly forecasts from a fast-moving consumer goods company, the analysis identifies substantial positive autocorrelation, departures from normality, systematic forecast optimism, and heterogeneous relationships between forecast volume and error variance. The proposed method integrates several corrections into a modular, sequential safety-stock architecture. It constructs cumulative forecast errors over the operational action period, selects an SKU-specific parametric distribution, and determines a baseline protection point using an expected-shortage condition consistent with the target beta fill rate. A forecast-scaled module adjusts dispersion when error variance depends on forecast volume, while a Cycle-End Exposure safeguard limits within-cycle shortage risk when large replenishment lots increase exposure. Performance is evaluated through discrete-event simulation against the company's current method. The proposed method reduces average inventory by 9.9% and maximum inventory by 19.5%, with no statistically significant deterioration in beta fill rate or average stockout duration. The proportion of weeks without stockouts declines slightly, without evidence of more persistent or operationally severe shortages. The heteroscedasticity module provides selective benefits, with greater service improvements as variance dependence intensifies. The resulting framework is interpretable, adapts to SKU-specific statistical profiles, links distributional features to safety-stock outcomes, and can be integrated into existing ERP environments.

We determine the minimal stretch factors among all orientation-reversing pseudo-Anosov mapping classes of the closed orientable surfaces of genus two, three and four.

LLMs absorb vast knowledge from training data, yet their inherently probabilistic nature offers no guarantee that this knowledge will surface in their outputs. To bridge this gap, Knowledge Extraction Enhancement (KEE) techniques such as activation steering, decoding adjustments, and prompting modify inference, potentially reinforcing existing predictions or surfacing less accessible answers. Yet their evaluations centered on aggregate accuracy, which obscures diverse effects by collapsing recovered failures and disrupted successes into a single score. We introduce MAPLE, a unified framework for characterizing how KEE techniques affect parametric knowledge extraction. Its diagnostic metrics disentangle these effects at the surface, probability, and beyond-sampling levels by measuring answer recovery and disruption, gold-token rank shifts on unrecovered failures, and recovery of answers missed by fixed-budget sampling. We evaluate six KEE techniques spanning decoding, activation steering, and prompting across three model families: OLMo2, Qwen2.5, and Llama3. Our analysis reveals distinct effects and trade-offs across techniques, highlighting the value of MAPLE for understanding and comparing KEE beyond aggregate accuracy.

This paper proposes a Full Duplex (FD) framework for safeguarding Integrated Sensing and Communications (ISAC) against a passive multi-antenna Eavesdropper (Eve). The adversary is modeled as both a communication and sensing threat, seeking to decode uplink data streams and estimate the surrounding scene. Eve's threat is quantified through two worst-case requirements over a location uncertainty region: a minimum secrecy rate for data confidentiality and a minimum eavesdropping Cramer-Rao Bound (CRB) for sensing privacy. We derive the eavesdropping information measures governing both data interception and scene inference, showcasing that transmitting Artificial Noise (AN) can degrade Eve's capabilities through the inverse interference-plus-noise covariance. Exploiting the underlying structure of the problem, we characterize finite-power local and robust feasibility conditions for target secrecy-privacy pairs. The analysis showcases that AN is fundamentally subspace-limited: increasing power is beneficial only along Eve's communication channel and sensing-derivative directions illuminated by the effective AN subspace. Finally, we formulate an AN power allocation problem that is solved via monotone branch-and-bound, and numerical evaluations quantify the AN power required to guarantee joint secrecy and sensing privacy.

Integrating world models into Vision-Language-Action models (VLAs) has emerged as a promising direction, enabling policies to anticipate future dynamics beyond reactive observation-to-action mapping. However, existing approaches vary simultaneously in their world representations, integration paradigms, and policy backbones, making it difficult to determine what truly matters for effective world model integration. To disentangle these factors, we categorize existing efforts into four representative paradigms, namely Implicit Future Alignment, Explicit Future Query, MoT Future Expert, and External World Predictor. We instantiate each within a unified framework and systematically evaluate these paradigms with diverse world representations and policy backbones across in-domain, out-of-domain, long-horizon, dynamic, few-shot, and real-world settings. Our study yields four main findings: 1) world representation effectiveness is paradigm-dependent, whereas 2D semantic representations offer robust, cross-paradigm gains; 2) no single paradigm is universally optimal, yet the decoupled MoT Future Expert offers competitive overall performance, particularly in OOD robustness; 3) world modeling consistently benefits policies built on standard VLMs, while decoupled designs exhibit greater compatibility with pretrained VLAs; and 4) world modeling improves performance over action-only baselines across multiple settings, with notable gains on long-horizon and dynamic tasks. Project page and video demonstrations are available at https://dylanmqs.github.io/WM-in-VLAs/.

This paper studies resilient angle-constrained leader–follower formation control and localization of multi-agent systems under asynchronous denial-of-service (DoS) attacks, which block individual sensing and communication links. To preserve the well-posedness of angle-based localization, an all-or-nothing triangle integrity protocol (ANTIP) is proposed to convert edge-wise attacks into triangle-wise ones, and DoS attacks are classified into localizability-maintained and localizability-broken attacks. First, with relative position measurements over attack-free links, resilient formation tracking laws are designed for leaders with constant velocities and with time-varying velocities comprising translational, rotational, and scaling motions. These laws guarantee asymptotic tracking and finite-time recovery of tracking, respectively, when the duration of localizability-broken attacks is suitably bounded. Second, with only angle measurements and communications, a resilient simultaneous localization and formation control (RSLAF) scheme is developed. It combines a finite-time distributed observer, which estimates the exosystem matrices and states under asynchronous DoS attacks, with a perturbation-based localization law that prevents localization from getting stuck at degenerate configurations. The theoretical results are rigorously proved and illustrated by numerical simulations.

Text-to-video (T2V) generation models can produce visually realistic videos that nevertheless violate the physical consequences implied by the prompt and scene. Existing benchmarks largely rely on holistic plausibility scores or broad quality dimensions, leaving it unclear which physical behavior is being tested and why a model fails. We introduce PhyMetric, a scene-level diagnostic benchmark that recasts physical plausibility evaluation as verifiable, scene-grounded question answering. PhyMetric contains 1{,}115 scenes spanning 50 physical categories across six domains, paired with 8{,}920 answer-balanced binary questions over 12 evaluation dimensions. Each scene is associated with eight targeted yes/no questions and a structured physical prior specifying the governing law, expected temporal sequence, observable signatures, and common failure modes. At evaluation time, a vision-language model (VLM) judge answers each question from question-aware sampled frames conditioned on the prior, and correctness is hierarchically aggregated into scene, dimension, physical category, and model scores. Experiments on four open-source T2V models and three VLM judges show that scene-specific binary QA is a critical component of the protocol: when each targeted question is replaced by a single generic plausibility query, evaluator accuracy drops to the chance level. Structured physical priors offer a complementary signal that improves human alignment of the resulting scores, with effects that vary across VLM judges of different capacity. These findings establish PhyMetric as a benchmark that turns physical plausibility evaluation from opaque holistic scoring into interpretable, scene-level inspection. Our project page is available at: https://mengsiwei.github.io/PhyMetric/.

LLM agents are evolving from a single model calling tools to planners that orchestrate multiple expert models whose capabilities live in distinct weights. This shift breaks a hidden assumption of today's serving systems: that the working set of models fits in GPU memory. On the memory-constrained GPUs typical of private and edge deployments, naive whole-model swapping serializes weight transfer with computation and, without host-backed KV restoration, discards session state at a switch. We present JointServe, a serving system that treats weight shards and KV blocks as residency objects competing for one unified memory budget. JointServe exploits three structural properties of agent workloads (directed calls, gapped execution, and dual-state residency) to prepare models before they are called: an intent signal parsed from the caller's own output stream triggers prefetch during generation, partial k* residency reduces switching to incremental layers, and a preemptive transfer scheduler protects live-session KV with a guaranteed bandwidth share. A carve-out control step adjusts the weight/KV allocation as workload demands change. Evaluations show that JointServe reduces session completion time by 21.7% over the strongest whole-model swap baseline and stays usable in a crash regime where every whole-model swap baseline fails to finish (a 46-fold gap). On replayed SWE-bench, tau-bench, and OSWorld traces, it cuts session completion time by up to 49% and lifts SLO attainment from 2.8-24.5% to 76.8-100%.

Repeated high-fidelity state and gradient evaluations make large-scale PDE-constrained optimization computationally expensive. Reduced-order modeling provides a natural way to reduce the cost of these repeated high-dimensional computations. However, existing reduced-order modeling approaches often rely on converged high-fidelity states for model construction or updating, while their reduced spaces may lose accuracy as the design moves beyond the region represented by the available data. To address these limitations, we present Reduced-order Trajectory-based Optimization (RETO), an online framework that constructs local reduced-order models (ROMs) directly from unconverged primal solver trajectories. Each local ROM is built from snapshots collected during a finite number of high-fidelity primal iterations, without requiring full-order convergence, and supports multiple low-cost state evaluations, adjoint-based gradient evaluations, and design updates. New solver trajectories are generated at updated design points to reconstruct the local ROM and continue the optimization. Across three distinct PDE-constrained optimization problems, RETO achieves solution quality comparable to that of full-order optimization, with speedups of approximately 8 to 22 times in cases without discontinuities in the state fields. Even in the transonic inverse-design case involving moving shocks, RETO retains an approximately threefold speedup. The transonic results further show that the online local ROM progressively captures evolving shock-related flow features as new primal solver trajectories are incorporated. These results demonstrate the value of unconverged primal solver trajectories as an online information source for reduced-order optimization, allowing the reduced basis to evolve with the design rather than requiring a basis that covers the entire design space in advance.

Distilling capable language-model agents into smaller models requires aligning the student's behavior with the teacher's. This raises a fundamental question: at which states and at what granularity should this alignment be performed? Offline imitation provides dense but token-specific supervision and can suffer from state-distribution mismatch. Token-level on-policy distillation (OPD) addresses this mismatch but typically requires teacher token probabilities and compatible tokenization. We argue that semantic actions provide a natural unit of supervision for agent distillation. We introduce ActionOPD, an action-level on-policy distillation method motivated by maximizing expected teacher-action likelihood at student-encountered states. At each such state, ActionOPD samples actions from both the student and a black-box teacher, rewarding student actions according to their semantic agreement with the sampled teacher actions. Through policy optimization, ActionOPD aligns the student's action-level behavior with the teacher's decisions without requiring reproduction of teacher-written reasoning or access to teacher token probabilities. Training Qwen3-4B and Qwen3-8B with two teachers on diverse, curated tool-use tasks yields average gains of 3.3-6.9 percentage points across four out-of-distribution benchmarks. ActionOPD scales favorably with student size and teacher capability, without requiring the teacher and student to share a model family or tokenizer. On $\tau^2$-bench, ActionOPD achieves competitive performance against offline distillation and token-level OPD. We further explore Self-ActionOPD with a same-model teacher and privileged context. These results establish ActionOPD as a promising approach to agent distillation.

The rapid adoption of Retrieval-Augmented Generation (RAG) and long-context LLMs has shifted the computational burden toward the prefill phase—the initial stage where the model processes all input tokens in parallel before generating a response—yet its energy impact remains poorly characterised. We investigate the contradiction between massive hardware-level power spikes and the sub-linear marginal cost of input tokens reported in system-level studies. Through high-resolution telemetry of Llama-3.1 and Qwen-2 models on NVIDIA A100 hardware, we identify the efficiency threshold where $O(n^2)$ attention complexity overcomes GPU parallelisation. Our results quantify how optimised memory orchestration (PagedAttention) delays this scaling bottleneck across three orders of context scaling, yielding up to an 8.9$\times$ energy efficiency advantage at 112,000 tokens compared to standard baselines. Furthermore, we characterise the energy disparity between input and output tokens, demonstrating that optimised serving (vLLM) causes this ratio to narrow (from $\approx$850$\times$ to 750$\times$) as context scales due to high decode efficiency, whereas unoptimised baselines show a much smaller, roughly constant gap ($\approx$150–230$\times$) because their prefill is itself expensive. This work establishes context-scaling laws that differentiate engine-level optimisations from inherent model constraints, providing an empirical framework for sustainable LLM deployment.

Agentic AI systems combine model decisions, mutable state, tool calls, and delegation to external agents. Monitoring them requires observations from different execution boundaries, while prevention requires control before an operation produces its effect. We present AgenticRV, a runtime-verification architecture that separates these boundaries from the monitoring formalism. Adapters emit canonical events; each monitor receives its own projection and abstraction; and a decision policy uses normalised verdicts to select a reaction supported by the responsible adapter. We formalise when observations suffice, monitor replacement preserves decisions, and blocking prevents unsafe effects. We also distinguish rejected attempts from completed effects, avoiding the assumption that a blocked request must leave every monitor in a permanent violation state. The Python implementation supports LangGraph, MCP, and A2A. A controlled LangGraph experiment combines an external Reelay temporal monitor with a programmatic report monitor; ablations show why both are needed for the selected requirements. Differential checks and cross-adapter tests provide further integration evidence. An LLM-driven incident-response example illustrates the same enforcement path, while existing microbenchmarks isolate framework costs. The results support reuse of monitoring methods at explicit control boundaries, subject to the stated observation and deployment assumptions.

Human values shape how we expect language models to behave. Dominant alignment approaches treat values as global objectives, although the appropriate value orientation is personal and contextual. Training-free alignment can steer models toward target values at inference time, yet has been largely evaluated in single-turn settings and with a single steering direction (e.g., toward safety). We evaluate four prompt- and activation-based steering methods in multi-turn conversations, where user and assistant models push toward the same or opposing, equally desirable values. We track alignment with a linear probe across three models and two value trade-offs. When assistant and user pull in opposite directions, the side matching the assistant's baseline bias often prevails, and effects that look successful early often fade or reverse after a few turns. Steering should therefore be evaluated over full conversations and in both directions of a value trade-off.

Resonantly driven nonlinear systems, from superconducting circuits to nanomechanical resonators, are often described in a rotating frame, where amplitude and phase evolve slowly within a fixed landscape of stable and unstable states. How such systems behave when this landscape moves on their intrinsic timescale remains to be understood. Here, we exploit the strong tension sensitivity of a graphene drum's resonance frequency to reshape the rotating-frame phase space of a bistable mechanical mode in real time, combining an MHz-frequency resonant drive with kHz-frequency tension modulation. When the modulation rate becomes comparable to the mechanical relaxation rate, adiabatic tracking of the fixed points breaks down. Instead, parametric instabilities emerge in the slow dynamics at frequencies set by libration around these points rather than by the mechanical resonance; stronger modulation drives period doubling and chaos. A slow-variable theory predicts where the instabilities occur, and Melnikov analysis captures the onset of chaos. These results show that the rotating frame itself can be parametrically driven, offering a general route to controlling nonadiabatic dynamics in systems ranging from Josephson bifurcation amplifiers to cold atoms.

Dynamical system (DS)-based learning from demonstration generates flexible robot motions, but existing guarantees rely on certificate generation to obtain formal guarantees of safety and goal convergence. We propose a symbolic supervisory control architecture that automatically generates a fail-safe backup plan, certifying safety, reachability, and actuator feasibility directly from the robot's dynamics, independent of the learned DS. A nominal controller tracks the learned DS, while the supervisory controller monitors safety and target convergence, providing non-conservative intervention when the nominal controller violates necessary constraints. Experimental results include simulations and real robot deployment with various 2D and 3D datasets, including handwriting datasets, validating the effectiveness of our approach in providing formal guarantees with learned motion even in complex, dynamic environments.

Transformer-based tabular foundation models (TFMs) achieve state-of-the-art performance on small to medium-sized tabular prediction tasks while processing diverse inputs through a shared stack of layers. Recent layerwise analyses identify strong predictive performance in early TFM representations and substantial overlap in computation across layers. However, early predictive accessibility alone does not establish whether later layers are dispensable for every input, and aggregate accuracy can obscure which inputs benefit from additional computation. We study all 24 in-context-learning layers of TabPFN-3 across 272 classification datasets, combining representation geometry, prediction trajectories, truncation, and structural interventions, while also examining related representation dynamics in four additional TFMs. We uncover a pronounced expansion-compression trajectory, with stronger late dimensional contraction associated with larger increases in class-prototype separation. A targeted analysis of early-stabilizing samples shows that substantial representational changes persist after predicted classes stabilize, while locally ambiguous queries obtain larger accuracy gains from shallow to late layers than well-separated queries. Repeating a single pretrained block or perturbing layer order reduces predictive performance on average, constraining simple post-hoc substitutions in the frozen model. These findings distinguish predictive accessibility, class stability, and the benefits of continuing computation, supporting an input-dependent assessment of depthwise redundancy in TFMs.

Spatio-temporal forecasting is fundamental to many real-world systems, yet deployment environments are inherently varying: temporal distributions may evolve, spatial domains may change, and observation availability may shift from training conditions. Existing models are often tightly coupled to their training environments through node-specific representations, predefined spatial support sets, and reliance on consistently available observations, limiting generalization to changing environments. To this end, we propose VaryST, a compact spatio-temporal forecasting model trained from scratch that achieves strong generalization by dynamically constructing predictive context from available long-term observations. To accommodate temporal changes, VaryST derives node representations from long-term histories, capturing current behavior in the context of longer-term patterns. To adapt to spatial changes, it dynamically selects functionally relevant support nodes beyond predefined neighborhoods and integrates their information with recent and long-term target states and global context. To handle varying observation availability, VaryST selectively masks historical patches on which the model relies most during training, encouraging it to exploit alternative temporal and spatial dependencies for accurate forecasting when key observations become unavailable. Extensive experiments on six real-world datasets against seventeen baselines show that VaryST achieves strong in-domain forecasting accuracy, substantially improves zero- and few-shot generalization under spatial and temporal changes, and remains robust to varying observation availability.

Let $f(z,w)=(p(z),q(z,w))$ be a polynomial skew product of $\mathbb{C}^2$ of degree $d\geqslant 2$ and let $\mu$ and $\mu_p$ be the Green measures associated to $f$ and $p$ respectively. Let $\lambda_1 > \lambda_2$ and $\lambda_p$ be the Lyapunov exponents of $(f,\mu)$ and $(p,\mu_p)$ respectively. We show that if $\lambda_2 = \lambda_p$, then $\mu$ is exact dimensional and: \begin{equation*}\dim_H\mu =\frac{\log d}{\lambda_1}+\frac{\log d}{\lambda_2}.\end{equation*} The proof relies on the continuity of the conditional measures obtained by Jonsson, on the fact that $\mu$ has constant Jacobian $d^2$, and on a fibered version of a normal form theorem for the inverse branches of the iterates of $f$.

LLM workflows and agents rely on a set of binary decisions like allow/block an input, or discard/retain a retrieved document. When human deferral is unavailable, uncertain cases must still receive an action. We study System One decision scores in this setting by defaulting to a conservative fallback when uncertain, and controlling the false omission risk within the subset of inputs assigned the utility-maximizing action. Across six datasets covering moderation, safety, and reranking, we compare general System One models with task baselines and a local open-weights fine-tuned System One model. We show that naive baselines for threshold selection exceed the target risk in roughly half of settings with defined test risk, compared with 7.8% for Learn Then Test (LTT), which leverages finite-sample statistical guarantees. Our results show why relying on a probability-like interface is insufficient for deployment decisions and how risk-budgeted selection trades useful action against observed error.

Large language models are increasingly competent at leveraging scientific knowledge and existing computational tools to execute complex workflows. At present, little information exists on their ability to reconstruct tools when given limited descriptions of such tools. We introduce an evaluation titled ReproBAIT (Reproducing Biological AI Tools), which measures the ability of coding agents to design, implement, and train biological AI tools taken from published research across a range of information access conditions. The coding agent is securely sandboxed without internet access and receives training data, a fixed time and compute budget, and limited information about the tool to replicate. We then score the reproduced biological AI tool based on its performance (tool-specific metrics) on held-out data. We evaluated a set of biological AI tools of varying complexities: ProteinMPNN, an inverse protein folding tool, RFdiffusion3, an all-atom generative diffusion tool, and a closed-source molecular design tool. We institute a methodology to check for memorisation, as the open source tools are likely to be included in models' pretraining corpus. Across a range of leading closed- and open-source coding agents, we find substantial reproduction capability for some selected tools, though limited capability for a highly complex tool. We find moderate evidence consistent with memorisation, varying across coding agents. These results suggest that tool developers need to be increasingly cautious when releasing information about the tools publicly; re-creation of the tool based on such information could undermine a tool's commercial value or invalidate dual-use managed access strategies.

Layer polarization provides a means of tuning the competition between interlayer and intralayer pairing in bilayer nickelates. Using ground-state constrained-path quantum Monte Carlo, we study interlayer $s^{\pm}$ and intralayer $d$ pairing in a bilayer two-orbital Hubbard model with a layer potential-energy difference $\varepsilon$ representing an effective perpendicular electric field. Within the parameter range studied, increasing $\varepsilon$ generally enhances the finite-size pairing vertex contribution $V_d$ and reduces $V_{s^{\pm}}$, shifting the relative pairing preference toward the $d$-wave channel. Increasing the onsite repulsion enhances both vertex contributions, more strongly in the $d$-wave channel, whereas increasing the crystal-field splitting reduces both and makes $s^{\pm}$ pairing relatively more competitive. Hund's coupling increases the relative competitiveness of $s^{\pm}$ pairing at weak bias, while enhancing the relative strength of $d$-wave pairing at large bias. The bias primarily redistributes charge between layers within each orbital sector, accompanied by an increase in the antiferromagnetic spin structure factor. Hund's coupling also enhances local interorbital parallel-spin correlations. These observations suggest possible connections between charge redistribution, magnetic correlations, and pairing competition, offering insight into how layer polarization and multiorbital interactions influence the relative pairing tendencies in the bilayer nickelate model.

Network Intrusion Detection Systems (NIDS) based on deep learning often fail to detect attack classes that are absent from training data, limiting their robustness against zero-day threats. This property called transferability across attack classes, remains underexplored especially in federated learning (FL) settings where data heterogeneity and client drift pose additional challenges. Prior work (TabFIDS) showed that FedAvg combined with bootstrapping and temporal averaging improves transferability for cross-device NIDS. In this paper, we explore whether other federated optimizers can do better. We compare two optimizers inspired by FedAdam and FedProx against the FedAvg baseline on the CIC-IDS 2017 dataset using an 11-client federated setup, sweep 10 hyperparameter configurations, and contrast two data-balancing strategies: bootstrapping and SMOTE. Using the count of client pairs whose transferred-model accuracy exceeds 70% as our primary metric (following TabFIDS), the proximal optimizer with a strong regularization term yields 48 transferable pairs versus 42 for FedAvg, while the adaptive optimizer with an aggressive server learning rate yields 46 pairs. SMOTE improves per-pair accuracy over bootstrapping but reduces breadth, revealing a depth-vs-breadth trade-off: high accuracy on a few client pairs versus moderate accuracy across most. These results indicate that the choice of federated optimizer, the rule by which client updates are regularized and aggregated into the global model, is very important for federated NIDS.

We prove that a bounded pseudoconvex domain with piecewise $C^2$-smooth boundary which has a finite volume quotient for the complete Kähler-Einstein metric is biholomorphic to a product of complex unit balls. More generally, the same conclusion holds for a bounded domain possessing a boundary point satisfying Kodama's conditions (C.1)-(C.5), provided the quotient has finite volume for a complete $\operatorname{Aut}(D)$-invariant Kähler metric. For piecewise Levi-flat boundary the product is a polydisk and in the smooth pseudoconvex case it is a single ball.

Among the various space interpolation operators used in high-order semi-Lagrangian schemes, the case of finite element interpolation presents a somewhat incomplete stability theory, partly due to the impossibility of applying in a standard form the framework of Von Neumann analysis, even in the constant coefficient case. In this work, the stability of the P_2 semi-Lagrangian scheme is stated in the case of the constant coefficient equation on a uniform mesh, using a suitable extension of the Von Neumann technique to treat block circulant matrices.

Densities and excess molar volumes ($V_{\text{m}}^{\text{E}}$) at (293.15 to 303.15) K and excess molar enthalpies ($H_{\text{m}}^{\text{E}}$) at 298.15 K are reported for the 2-butanol + 2,5-dioxahexane, or + 2,5,8-trioxanonane, or + 2,5,8,11-tetraoxadodecane, or + 2,5,8,11,14-pentaoxapentadecane systems at 95 kPa. Densities were measured using a densitometer Anton Paar DSA 5000, and excess enthalpies by means of a micro-calorimeter Tian-Calvet. $H_{\text{m}}^{\text{E}}$ and $V_{\text{m}}^{\text{E}}$ data are positive, indicating that interactions between like molecules are dominant and that the main contribution to $V_{\text{m}}^{\text{E}}$ is the interactional one. Our enthalpic data together with data from the literature have been used to evaluate the enthalpy of alcohol-glyme interactions, which are ranged between ($-$31 and $-$22) kJ mol$^{-1}$. From the comparison with $H_{\text{m}}^{\text{E}}$ results for glyme + heptane systems, and from the position of the investigated mixtures in the diagram $G_{\text{m}}^{\text{E}}$ (excess molar Gibbs energy) vs. $H_{\text{m}}^{\text{E}}$, it is shown that interactions are of dispersive type. 2-Alkanol + dipropyl ether (DPE), or + glyme systems have also been studied using the DISQUAC, UNIFAC (Dortmund) and ERAS models. According to ERAS results, physical interactions in glyme systems are dominant, while association effects are predominant in mixtures with DPE. The DISQUAC interaction parameters suggest that interactions related to the OH/O contacts are essentially dispersive. UNIFAC fails when representing thermodynamic properties of the studied solutions.

We study finite stochastic partial monitoring (PM) games with side information, a paradigm of sequential learning with partial feedback. On each round $t \in [T]$, both player and environment observe a context and player chooses an action and environment simultaneously chooses an outcome. The player neither observes the outcome nor the loss; instead, they only observe a feedback symbol which is partially informative about the outcome. We utilize the Estimation-to-Decisions (E2D) meta-algorithm of Foster et al.(2021) to design E2D.PMSide, a computationally efficient framework for contextual partial monitoring, which uses an online regression oracle to efficiently estimate the outcome distributions based on historical (context, action, feedback symbol) tuples. Since the exploration rule given by the original E2D algorithm is intractable to compute, we propose two computationally efficient exploration rules that achieve optimal regret orders in $T$: in the globally observable setting, our exploration rule is based on $\epsilon$-greedy, while in the locally observable setting, our exploration rule uses inverse gap weighting (Foster & Rakhlin, 2020), which was previously only known to be effective in the bandit setting. We show that a variant of E2D.PMSide named OE2D.PMSide can be combined with offline regression oracles to give an efficient algorithm with optimal regret when the contexts are drawn independently from a fixed distribution. We evaluate our algorithm on a synthetic label-efficient prediction game and an LLM routing task, in which a learner decides from prompt, uncertainty, and internal state features whether to escalate a query from a weak model to a strong one. We show that our method has superior performance to prior approaches.

Active domain adaptation researches focus on alleviating the annotation cost, which comprises source data and target data labelling, as well as enhancing the performance of unsupervised domain adaptation. However, optimizing the adaptation model based on the average accuracy can lead to the drop in performance of some difficult categories. To overcome this issue, a point-wise selection strategy built upon the influence function can be used to improve domain adaptation performance. In our work, we first address the sensitive and redundant aspects of point-wise selection, which can lead to overfitting of the acquisition function. We thus introduce a region-based approach that ensures the improvement of the adaptation model by maximizing a positive manifold coverage. An uncertainty score is further introduced to the selection mechanism to learn a better decision boundary for the target domain. In comprehensive experiments, our proposed method surpasses the first category-aware active domain adaptation baseline as well as existing active domain adaptation methods across multiple domain adaptation benchmarks. Our implementation is available at: https://github.com/nhtm2702/WHO IS NEXT.

Negative guidance of diffusion models generates samples that avoid an undesired condition, specified either by reference samples or by a conditional score alongside an unconditional model. Existing methods inject repulsive terms derived from local information at each sample, so the resulting sampler does not target an explicit density, and its bias cannot be systematically reduced by increasing computation. We instead present a negative-guidance method that samples from an explicit target density, which uniformly handles both reference-sample and score-based specifications and suppresses probability mass near the undesired condition with an avoidance strength that parametrizes the target density itself. Realizing this target requires estimating, at each diffusion timestep, how likely each sample is to belong to the undesired condition. We cast this estimation as a Positive-Unlabeled learning problem that admits lightweight online training along the diffusion trajectory, using condition samples as positives and diffusion particles as unlabeled data. Because the guided particles drift from the unconditional marginal, we further correct the residual mismatch via Sequential Monte Carlo, yielding a sampler that is asymptotically exact in the particle limit and that turns any existing negative-guidance method into its proposal kernel. Empirically, our method approaches the ground-truth target on Gaussian mixtures and Pareto-dominates existing methods on trade-offs between avoidance and generation quality, distributional bias, and diversity, including in an image-generation setting with a pretrained score network.