Papers

We consider steady-state optimization of nonlinear dynamical systems under time-varying stochastic operating conditions, where the objective is an expected-value function over the current operating-condition distribution, so that changes in the distribution induce a moving stochastic optimum. To address this problem, we propose stochastic sequential feedback optimization (S2FO), which combines projected feedback-gradient updates with sequential sensitivity estimates obtained by linearizing the plant at realized stochastic operating conditions. We analyze the closed-loop performance of S2FO through expected dynamic regret, instantaneous tracking in expectation, and almost-sure convergence under diminishing step sizes. We then apply the method to wind farm power maximization under time-varying stochastic inflow. Simulations on a benchmark layout and on a realistic offshore wind farm show that S2FO improves average power generation when persistent wake interactions are present, and that its advantage decreases as wind-direction variability increases.

Recent advances in robot learning have enabled generalist control policies capable of completing a wide range of tasks. However, their performance degrades when deployed in unseen environments, making it critical to detect failures and teach recovery behaviors. Existing runtime monitoring methods often require task- and policy-specific training or hyperparameter tuning, limiting cross-task deployment and introducing additional overhead during iterative policy updates. We present Reward-DAgger, a robot-gated interactive imitation learning framework that uses dense progress signals from a general-purpose reward model to determine when human intervention is needed. Our approach is agnostic to the underlying policy architecture, requires no access to policy internals, and can be applied across tasks without retuning the gating mechanism. Our results show that Reward-DAgger achieves a better failure-detection accuracy-latency tradeoff than existing runtime monitoring baselines. Across eight simulated and real-world tasks, Reward-DAgger consistently improves the downstream policy's autonomous success rate throughout interactive learning and achieves strong return on human effort, outperforming the baselines in most settings. Importantly, the same gating configuration is used across tasks without task-specific hyperparameter tuning, demonstrating transfer across tasks, environments, and policy architectures. Code and videos are available at https://liralab.usc.edu/reward-dagger.

Autonomous agents built on Large Language Models (LLMs) need standardized protocols to interoperate across systems. Several now exist (A2A, MCP, ACP, ANP, NLIP), but the Natural Language Interaction Protocol (NLIP) has not appeared in any controlled performance study, and no work has measured where an agent protocol's latency is spent. We compare NLIP and the Agent-to-Agent (A2A) protocol empirically, decomposing latency into message creation, connection, and send phases across three independent hardware environments. For lightweight coordination, NLIP is 8.4-9.6x faster than the baseline A2A SDK implementation on two environments and about 4x on a third; the direction of the advantage is consistent, its magnitude depends on the hardware. The advantage is stage-specific: for the end-to-end pipeline, where LLM inference dominates, the protocols are near parity. The difference comes almost entirely from connection setup. To test A2A at its best, we also ran A2A SDK with connection caching enabled; caching narrows its gap with NLIP by a hardware-dependent amount, from 2.75x on one machine to near-parity on faster hardware, where at scale a cache-optimized A2A-SDK matches NLIP. We report these as measured conditions without a single causal account of the residual send-phase cost. Against the more optimized Python-A2A, NLIP leads by about 4x on the same stage. We close with a protocol-selection guide keyed to workload characteristics.

Shape from shading using a single image of a Lam- bertian surface is inherently ambiguous. When the light source direction is known, the surface normal estimation has a cone- ambiguity, which worsens when the source is unknown. Recently, shape from heat conduction has emerged as an approach that leverages heat transport equations to estimate the Shape Lapla- cian operator, an intrinsic measure of shape. However, deriving surface normals from the Laplacian operator encounters a local binary convex/concave ambiguity. Our contribution introduces a novel theory to resolve these local shape ambiguities (excluding a few degeneracies) without relying on priors like smoothness, by combining the cues from shading and heat conduction. Our method ensures the mathematical constraints of both shading and the Laplacian are satisfied simultaneously, even with an unknown light source. We validate our theory through simulations of complex shapes and analyze its performance in the presence of noise, as well as on a noisy single thermal video of real-world objects with complex shapes and material properties, including varying albedo.

Accurate early screening for Autism Spectrum Disorder (ASD) is a precursor to timely intervention, which is critical for improving cognitive and behavioral outcomes. We present ASD-FEAT (ASD - Feature Extraction And Tracking), a multimodal dataset derived from video recordings of infant-adult interaction sessions. The key contribution of ASD-FEAT is the combination of longitudinal coverage from infancy through 36 months, repeated interaction sessions, clinically validated developmental outcomes, expert frame-level behavioral annotations, and privacy-conscious multimodal feature representations. To the best of our knowledge, existing ASD behavioral datasets do not jointly provide these characteristics at comparable scale. To demonstrate the utility of ASD-FEAT, we use it to evaluate a computer-vision-based end-to-end pipeline relying on machine learning techniques to automatically identify ASD risk. ASD-FEAT integrates both expert-defined and deep-learned features, including face and eye landmarks, facial action units, gaze direction, head position, mel-spectrogram audio representations, and optical flow, to identify behavioral markers of social interaction. Our automated pipeline achieves an ASD classification accuracy of 76.2% and an Area Under the Receiver Operating Characteristic (AUROC) of 0.82, compared to classifiers trained on manually labeled behaviors, which yielded 81.3% accuracy and an AUROC of 0.88. We further introduce a within-visit partner contrast: a per-visit signal contrasting examiner-directed and parent-directed social behavior which, when added to the classifier, lifts the fully automated Look Face + Smile and Look Face + Vocal configurations to Matthews correlations of 0.49 and 0.45 respectively, exceeding the human-coded single-partner baseline of 0.42.

We study nonnegative, non-monotone $k$-submodular maximization with $k\ge2$ labels under support constraints, and show how the certified approximation coefficient improves as the support region permits more uniform selection. For a compact convex down-closed support region $P\subseteq[0,1]^n$, the diagonal level $\zeta(P)=\max\{t\in[0,1]:t {\bf 1} \in P\}$ ranges from $\zeta=0$, which carries no geometric promise, to $\zeta=1$, which is unrestricted support. Our main structural result is a comparator-uniform linearization of the multilinear extension, built from an objective-independent action and a comparator-independent update field. For $k\ge3$, its validity reduces, independently of the number of elements, to four polynomial inequalities of degree at most three in one or two variables, only one of which depends on $k$. Explicit parameter choices give a nondecreasing certified profile $\underline{\alpha}_k(\zeta)$, in closed form on all of $[0,1]$ when $k=2$. At $\zeta=0$ we certify $0.4456\ldots$ for $k=2$ and $0.4541\ldots$ for every $k\ge3$, improving the recent $\sqrt2-1$ guarantee for one matroid or one knapsack, as well as the $1/3$-type guarantees for a fixed number of budgets; at $\zeta=1$ we certify $1/2$ for $k=2$, $(\sqrt{17}-3)/2$ for $k=3,4$, and $k/(2k-1)$ for $k\ge5$, whose excess over $1/2$ is of order $1/k$ rather than the previous $1/k^2$. Value-retaining rounding transfers these guarantees to matroid and knapsack constraints, and the same field yields $O(\sqrt T)$ approximate regret online under gradient or post-decision value feedback.

A language model reads long text in one quadratic forward pass, stops at the context window, and loses accuracy with length before reaching it. We ask whether the read can be factorized when deciding over a finite set: which document is relevant, which option is supported, which passage is the evidence. Periscope, a training-free inference method, arranges the $N$ chunks of a text on a $K{\times}K$ grid with $K{=}\lceil\sqrt{N}\rceil$ and asks a frozen model the same question about $K$ local spans of consecutive chunks and $K$ strided spans that sample the whole text, reading the log-odds of every answer at one token. Each answer takes its best local and strided score, and scoring every chunk by its two spans gives an evidence map at no further cost, whose peak is the chunk behind the answer. Every probe is about $\sqrt{sc}$ tokens for a text of $s$ tokens and chunk size $c$, so a window of $W$ tokens reaches $W^{2}/c$ tokens at $s^{1.5}$ cost. The map replaces the long read. On LongBench v2, reading only the $K$ chunks the map ranks highest, 9k tokens, matches the same model's best window read across windows from 32k to 1M tokens, and on InfiniteBench, where the median context is 150k tokens, it leads the best window read by 5 points. The same map ranks BRIGHT's long-document corpora with the best NDCG@10 of six methods. Each call caches only one probe, so a 27B model reads 4.5M-token contexts on one 80GB GPU, where a single pass would need 296GB of cache. A long read then needs a GPU that holds the model, not one that holds the text.

The quadratic attention cost of Vision Transformers (ViTs) forces a trade-off between spatial resolution and rollout horizon, particularly for fine-scale PDEs where shocks, reaction fronts, and material interfaces occupy small, evolving regions of the domain. Conventional neural surrogates also lack mechanisms to adapt resolution dynamically. We propose WAMRViT, a ViT that tokenizes inputs as balanced quadtrees using a wavelet-inspired refinement criterion, jointly encodes position and refinement level with 3D rotary positional embeddings, and regrids in cell space during inference for stable long-horizon rollouts. A multi-scale variant retains each leaf at its native source resolution and lets the model learn across resolution levels. Unlike fixed-budget adaptive-tokenization methods, WAMRViT imposes no predetermined token count and supports fully adaptive topology throughout autoregressive rollout. To our knowledge, it is the first machine-learning surrogate to natively tokenize multi-level Adaptive Mesh Refinement (AMR) data. On uniform-grid benchmarks, uniform-patch WAMRViT improves finest-level region-of-interest VRMSE over a finest-patch uniform ViT while using substantially fewer tokens. The multi-scale variant achieves the lowest first-step full-field RMSE and VRMSE on both benchmarks and improves rollout-averaged full-field and refined-region accuracy at long horizons. With parallelized regridding, its end-to-end rollout cost lies between finest-patch and approximately token-matched coarser-patch ViTs. On a complex AMR combustion problem whose finest features cannot be represented natively by the evaluated uniform-grid baselines, WAMRViT operates directly on adaptive cells and substantially reduces finest-level error at matched transformer capacity. Code: https://github.com/tonyzyl/wamrvit

Bipolar disorder is strongly associated with financial instability. We examine how different interventions motivate individuals with bipolar disorder to share financial data with others. This approach can inform the development of tools for digital monitoring and intervention designed to promote financial stability in this population. 500 individuals with BD completed a pre-registered factorial vignette survey to examine level of comfort with hypothetical scenarios involving third-party financial interventions during symptomatic and euthymic periods. Scenario components were systematically varied between third-party actors, mood states, and intervention types. Participants rated sharing comfort on a 0-10 point scale. Multilevel models tested differences alongside clinical and financial histories, relational trust, and personality. Participants were most comfortable involving care partners in financial planning. They were more comfortable with temporary spending restrictions during symptomatic states than euthymic periods, underscoring the importance of accurate mood detection for intervention delivery. Prior financial help-seeking behavior and higher relational trust predicted greater comfort. Bankruptcy experience — declared by 11.4% and considered by 31.7% — was associated with increased comfort with spending restrictions. Individuals with psychiatric advance directives (8%) were significantly more comfortable sharing spending behaviors than those without. Comfort with financial interventions was higher among those with prior financial challenges or help-seeking histories. Participants distinguished between symptomatic and euthymic periods, favoring targeted, time-limited restrictions over general monitoring. These findings extend prior work on financial data sharing for illness self-management.

Financial LLM agents are often evaluated by comparing their end-to-end returns with those of a baseline and testing the paired difference against zero. This measures whether deploying the agent changes realized performance, but it does not isolate event-selection skill. An agent that frequently changes positions from flat to long can earn a positive paired return from an upward-drifting event pool even when it selects events at random. We propose the Agent Policy-Value Audit, which holds fixed the observed count of each ordered action-change type and randomly reassigns them across eligible events. The average payoff from these reassignments is the composition benchmark; the difference between observed deployment value and this benchmark is selection value. In semi-synthetic benchmarks based on real earnings-event returns, a zero-centered paired test falsely attributes passive exposure to selection skill in $11.6\%$ of no-skill replications, while the transition-matched audit reduces this rate to $5.3\%$. Applied retrospectively to 723 earnings events at 44 U.S. consumer-facing firms, the audit decomposes the agent's gross deployment value of $+15.2$ bps/event into a $+25.8$ composition benchmark and a $-10.6$ selection value. The agent does not detectably outperform matched random assignments. Financial-agent evaluations should report deployment value separately from event-selection value.

Applications such as engineering design often require us to optimize a black-box function, i.e., a system whose inner processing is not analytically known and whose gradients are not available. Practitioners often have a fixed budget for the number of function evaluations and the performance of an optimization algorithm is measured by its simple regret. In this paper, we study the class of "Optimistic Optimization" algorithms for black-box optimization that use a partitioning scheme for the domain. We develop algorithms that learn a good partitioning scheme and use flexible surrogate models such as neural networks in the optimization procedure. For multi-index functions on an $m$-dimensional subspace within $d$ dimensions, our algorithm attains $\tilde{O}(n^{-\beta / d})$ regret, where $\beta = 1 + \frac{d-m}{2m-1}$, as opposed to $\tilde{O}(n^{-1/d})$ for SequOOL, a state-of-the-art optimistic optimization algorithm. We use our approach to improve the quality of Activation-aware Weight Quantization (AWQ) of the OPT-1.3B model, achieving $\sim10\%$ improvement in performance relative to the best possible unquantized model.

Synthetic data are increasingly used as an alternative to sharing sensitive records. However, synthetic data generation does not guarantee privacy, as diffusion models trained or adapted on sensitive data remain susceptible to reconstruction attacks. Moreover, while approaches that use differential privacy (DP), such as DP-SGD, achieve provably private diffusion model training, the repeated gradient clipping and noise injection they require result in significant utility loss. An important limitation of DP-based privacy is that, although it has a provable relationship to reconstruction privacy (RP), that relationship is indirect. RP is defined in terms of limiting how much an adversary's posterior distribution over sensitive data differs from the prior, whereas DP provides guarantees by bounding the sensitivity of outputs to changes in individual records. This indirection is an important source of the utility loss. To address this, we propose a PAC-private diffusion model adaptation to achieve reconstruction privacy. Since PAC-privacy is defined directly with respect to posterior advantage over the prior, it directly implicates RP. To obtain scalable PAC privatization in high dimensions, we first learn a compact data-dependent diffusion model component using LoRA or Textual Inversion, and then calibrate anisotropic Gaussian noise from the covariance of repeated mechanism outputs. Unlike DP-SGD, our method perturbs the learned component only once after optimization, thereby avoiding privacy composition across gradient updates. We evaluate the framework on few-shot concept personalization and full-dataset image synthesis, and show that the proposed approach better preserves subject identity, generation quality, and downstream classification accuracy than DP while achieving the same reconstruction privacy.

Modern GPU-based HPC systems rely on heterogeneous vendor stacks, yet resilience studies are largely limited to single architectures, leaving it unclear how faults behave across different GPU backends. We present BitIR, a cross-architecture fault injection framework that injects deterministic single-bit faults at the LLVM IR level, ensuring semantically equivalent perturbations prior to backend lowering and enabling direct cross-vendor comparison across NVIDIA, Intel, and AMD GPUs. We evaluate BitIR on three production supercomputers – Polaris (ALCF, NVIDIA A100), Aurora (ALCF, Intel GPU Max 1550), and Frontier (OLCF, AMD Instinct MI250X) – representing the full spectrum of current leadership-class GPU architectures. Using representative heterogeneous benchmarks, we conduct large-scale injection campaigns across all three systems and classify outcomes into masked results, silent data corruptions (SDCs), and failures. Our results show that identical faults produce markedly different behaviors across vendors: Intel most often masks faults but exhibits more hangs when faults escape masking, NVIDIA exposes more detectable hard failures, and AMD alternates between SDC-dominant and failure-dominant behavior depending on the benchmark and fault site. These findings demonstrate that resilience is not backend-invariant, underscoring the need for backend-aware and workload-aware fault mitigation.

Vision Transformers (ViTs) have shown strong performance in volumetric segmentation, but their effectiveness on clinical CT is limited by an isotropic Euclidean lattice assumption. This conflicts with anisotropic CT acquisition, leading to two key issues: (1) a metric mismatch between voxel indices and physical anatomy, and (2) accuracy degradation from isotropic resampling. To address this, we propose MAGEFormer, a geometry-calibrated framework that embeds physical metric constraints directly into representation learning. Our method introduces Metric-Adaptive Spatial Embedding (MASE) to calibrate positional frequencies using voxel spacing, Geometry-Constrained Attention (GCA) to suppress physically implausible feature correlations, and Geometric View Voting (GVV) to reduce discretization bias during inference. We evaluate MAGEFormer on two multi-organ abdominal CT benchmarks, BTCV and FLARE 22, under a unified protocol against strong CNN and Transformer-based baselines. MAGEFormer achieves the strongest boundary accuracy among the compared methods, with 10.58 mm HD95 on BTCV and 3.40 mm HD95 on FLARE 22, and shows consistent gains in Dice under the same protocol. These results show that geometry-aware internal calibration is more effective than relying on conventional isotropic preprocessing alone for anisotropic CT segmentation.

Data assimilation combines model forecasts with noisy, incomplete observations to estimate the evolving state of a dynamical system. Existing methods face two compounding challenges: high-dimensional nonlinear dynamics make repeated forward simulation computationally expensive, while sparse observations provide limited direct information about the full state. To address these challenges, we propose the Latent-Dynamics Ensemble Flow Filter (LD-EnFF), a sequential Bayesian filtering framework that performs both forecast propagation and filtering updates in a compact latent space. LD-EnFF combines a latent dynamics surrogate for ensemble propagation with a variational autoencoder (VAE)-based observation model that evaluates a state-dependent observation likelihood in latent space. At each assimilation step, an ensemble filtering update based on flow matching uses the forecast ensemble and this likelihood to generate posterior samples, jointly updating latent states and uncertain parameters. This design avoids repeated full-state simulation during forecasting and full-field reconstruction during likelihood evaluation. LD-EnFF substantially outperforms a broad range of data assimilation algorithms on benchmarks spanning Kolmogorov flow, tsunami propagation, and atmospheric modeling, all featuring complex dynamics and sparse, noisy observations.

Information and infectious diseases spread through social networks, but the spreading probabilities driving them are hard to estimate without per-node activation times. Applications seldom supply these, and inference must instead proceed through indirect and noisy proxies for the terminal infection states. We study this inverse problem on a fixed, known graph under the Hidden Independent Cascade (HIC) model, with one spreading probability per node rather than a single global rate, so the number of unknowns scales with the number of nodes. Seed sets and observation parameters are known, while activation times and terminal infection states are latent, and the observed-data likelihood requires marginalizing over every spreading outcome. We propose a simulation-based amortized estimator that recovers the full node-level parameter vector without reconstructing individual latent cascades. Repeated seed-conditioned symptom observations are summarized as Symptom-Aware Cascade Features (SACF), which combine empirical symptom statistics with neighborhood and structural information. SACF are mapped to parameters by SAGE-HC, a permutation-equivariant gated graph neural network whose learned gates attenuate neighborhood messages corrupted by false positives and false negatives, and training on simulated HIC realizations yields a reusable inverse map. As a benchmark under the same hidden observations, we extend the Dynamic Message Passing learning framework to the HIC emission model. On synthetic and empirical graphs the two methods separate by topology. DMP is highly accurate on trees, whereas SAGE-HC is substantially better on heterogeneous, loopy graphs under noisy terminal symptoms.

In autonomous mobility duopolies, platforms compete for riders based on vehicle proximity. A critical operational challenge—faced by an incumbent proactively defending against a market entrant—is how to dispatch fleets to minimize demand poached by a reactive competitor. We model this competition as a zero-sum robotaxi assignment game, focusing on computing optimal deterministic Stackelberg strategies for the leader. We formulate this problem as an exact single-level integer linear program and use its relaxation to design a polynomial-time approximation algorithm achieving an additive error from the optimum. We argue that this is not far from best possible by proving that deciding whether a strictly positive payoff is achievable is NP-complete, and that approximating the optimal payoff within an additive error scaling linearly with the number of riders is NP-hard. We also provide an exact polynomial-time dynamic programming algorithm for computing the optimal Stackelberg payoff when riders and vehicles are on the Euclidean line. Finally, we complement our theoretical findings with an empirical evaluation using real-world ride-hailing data. The experiments demonstrate that our LP-based algorithm significantly outperforms its worst-case guarantee in practice, while lightweight heuristics achieve highly competitive payoffs with negligible computational overhead, offering practical solutions for large-scale deployments.

Clinical MRI routinely acquires multiple contrast-weighted images of the same anatomy for complementary tissue characterization. However, current accelerated MRI methods typically reconstruct each contrast independently, without fully exploiting shared anatomical information. This work aims to learn anatomical representations invariant to contrast-dependent appearance for reconstruction of accelerated multi-contrast MRI. We propose MAX (MAnifold eXpansion), a subject-specific framework that learns anatomical representations from a single fully sampled reference contrast. To address the under-constrained separation of shared anatomy and contrast-dependent components from a single image, MAX expands the multi-contrast manifold using anatomy-preserving intensity augmentations. A disentangled implicit neural representation models augmented samples using shared spatial coordinates for anatomy and spatially invariant coordinates for contrast appearance. The learned anatomical representation is then fixed, with the contrast representation adapted to the undersampled target data, followed by unrolled refinement. Theoretical analyses further provide insight into the disentangled representation learning and explain how the learned anatomical representation improves the target contrast reconstruction. At R = 8 for brain MRI and R = 6 for knee MRI, MAX achieves the highest mean PSNR and SSIM across all tasks, improving PSNR by more than 1 dB over the strongest baseline for both brain contrasts. MAX more faithfully recovers subtle anatomical and pathological structures and remains robust to inter-contrast motion, structural heterogeneity between reference and target contrasts, and measurement noise. Therefore, MAX provides a general strategy for leveraging high-quality reference scans in accelerated MRI and has the potential to be extended to other reference-assisted MRI inverse problems.

Dense embedding models are core to modern text retrieval, enabling systems ranging from web search to retrieval-augmented generation (RAG). Yet, retrievers are usually deployed within opaque systems, exposing only ranked results, cited sources, or generated answers. This opacity prevents users from verifying which retrievers providers serve and may create a false sense of robustness against retrieval attacks. We show that attackers can infer retrievers' identity only through queries. We introduce TellTail, a fingerprinting attack for identifying retrievers behind black-box systems across a spectrum of access levels. Despite sharing many properties, retrievers can be steered to emit distinct results. When retrieved passages are exposed, TellTail compares retrieval overlap to deduce the underlying model. When only the final generated response is exposed, TellTail optimizes model-specific queries that induce a chosen retrieval behavior on the target retriever but transfer poorly to others. Interestingly, the poor transferability that limits attacks elsewhere is exactly what makes them useful for fingerprinting. We evaluate TellTail using 53 retrievers under three increasingly restrictive settings. TellTail perfectly identifies the deployed retriever from full retrieval rankings; in 94.3% of attempts with unordered top-3 results; and in 92.5% of cases from language-model-generated answers alone. Furthermore, fingerprinting succeeds against a popular RAG system (OpenWebUI). Overall, TellTail shows that retrieval-based systems leak the identity of their embedding model—exposing intellectual property and enabling reconnaissance ahead of model-specific attacks such as corpus poisoning.

Extreme values of a physical system are rare by definition, yet they can govern some of its most consequential and anomalous behavior. Their sparsity makes them particularly costly to identify in large data sets. Quantum computation offers a potential way to reduce this cost. The challenge is that quantum amplitudes cannot simply be read, copied, compared, or marked using conventional search oracles. Here we introduce an algorithm for searching amplitude-encoded quantum signals that exploits their underlying structure. Through a nonlinear amplitude transformation, the algorithm identifies and certifies a global extremum while exponentially reducing the effective search space for structured signals. We establish a polylogarithmic query complexity in the data size $N$, together with global completeness, efficient classical verification, and a measurable criterion for quantifying exploitable structure. We demonstrate the framework using computational turbulence data. Applied to line cuts of velocity gradients from a $1024^3$ simulation containing more than one billion grid points, the algorithm consistently isolates the extreme tail and reduces the median search space to $O(\log N)$. The clustered structure characteristic of turbulence further accelerates the search. These results establish a direct connection between physical structure and quantum advantage in extreme-value search. For accessibility, the main text emphasizes the central ideas and results, while the theoretical derivations are provided in the Appendices.

Powered neck exoskeletons are developed to restore head-neck motions for those with neck muscle weakness. Current methods using hand-held input devices to control these robotic systems are unintuitive or non-viable for these populations. Alternatively, the eyes coordinate with the head to accomplish many daily tasks. Therefore, movements of the eyes can be a useful signal to control the neck exoskeletons without needing to use the hands. However, previous control models lack biobehavioral basis as well as means to personalize the control behavior, causing unnatural head-neck movements. This paper presents a novel eye-tracking controller design leveraging the vestibulo-ocular reflex to emulate natural eyehead coordination. Personalization tools were also designed, consisting of sliders to allow users to adjust the parameters of this controller. Here, we present the technical design and validation of this system in human users. Compared to the previous eye-tracking model, the new controller provides natural head motions personalized to the user through more natural velocity profiles and a 24 percent average increase of accurate motion intent detection. This system provides a powerful platform for studying eye-tracking controllers in the context of personalization and preference, and ultimately restoring the lost head-neck mobility in patient populations.

Generative artificial intelligence has become increasingly incorporated into digital media and more generally into production workflows with which the public frequently interacts. Current provenance standards and disclosure methods frequently rely on binary categorizations, differentiating only between entirely human-authored and AI-generated content, or depend on technical watermarking that lacks user-facing clarity. However, there are very different risks and outcomes depending on the different uses of AI in the creation and consumption of end products. Therefore, we propose the Reported Engagement with AI Level (REAL) Rating framework which addresses these concerns by introducing a six-tier scale (Levels 0-5) that discloses the extent of AI involvement in both product creation and user experience. This paper details the structural methodology of the REAL Rating system and the specifics of its application across text, image, audio, video, and software products.

Sparse Autoencoders (SAEs) began as an unsupervised tool for decomposing neural representations into sparse, interpretable features, and are increasingly used not only for passive analysis but also for active interventions such as unlearning, bias mitigation, and concept editing. A central challenge for these editing and steering methods is reliably matching features to target concepts; most current approaches address this by computing post-hoc scores over an already-trained, frozen dictionary. We instead introduce ClasSAE, a novel method that both automatically assigns classes to features and guides the encoder toward class-separable representations during training. Specifically, we apply a differentiable top-$k$ operator to a trainable feature–class affinity matrix with per-feature budgets, coupling the features selected for each sample to the classes they are trained to represent. Because gradients flow through the selection of active features rather than only through their magnitudes, the encoder and the affinity matrix co-adapt rather than being fit in separate stages. The result is a dictionary that is both class-separable and class-annotated, with no need for post-hoc probing. We propose three variants for enforcing sparsity within this framework, which achieve comparable overall performance with slightly different trade-offs. Using CLIP ViT-L/14 embeddings on ImageNet, we show that the learned affinity matrix agrees closely with an independently estimated post-hoc feature-class matrix computed on held-out data. The model also supports direct class prediction from the encoder and affinity matrix alone, without a separately fitted classifier, and its more class-aligned encoder yields improved separation in Targeted Probe Perturbation evaluations. https://github.com/St0pien/ClasSAE.

Graph based cyber attack detection studies employ various graph construction and representation strategies across different cybersecurity application domains. This diversity motivates a quantitative examination of how representation strategies are distributed across these application domains. This study presents a quantitative analysis of 37 original studies published between 2019 and 2026. Each study was coded according to publication year, application domain, graph representation type, feature extraction strategy, learning paradigm, algorithm, and dataset. Frequency analysis, cross tabulation, and statistical association tests were applied. Automated representation learning was the most frequently employed strategy, accounting for 73.0% of the studies, while handcrafted representation strategies accounted for 27.0%. The Fisher Freeman Halton exact test revealed a statistically significant association between application domain and representation strategy (exact p = 0.008; Cramer's V = 0.540). The findings indicate that automated representation learning predominates within the analyzed corpus and that the distribution of representation strategies differs across cybersecurity application domains.

Localizing behavior to individual components of a language model is a central goal of mechanistic interpretability. However, scoring components one at a time misses context-dependent effects: a primary component can inhibit the activation of a backup, leading to issues with ranking components. Actual causality studies the structure of such interactions via witnesses: variables that provide contextual information to resolve interaction terms. However, estimation with witnesses typically requires combinatorial enumeration and is infeasible in practice. We introduce the witness-integrated set effect (WISE), a family of causal estimands that build on the witness mechanism while taking expectations over sets of causes and witnesses to remain computationally feasible. Building on this approach, we introduce JuntaLearner, a gradient-based circuit discovery method that learns to rank components by their causal impact across varying-sized sets of components and witnesses. Alongside faithfulness metrics, we introduce measures of necessity and task specificity, and the circuit recognition score (CRS) to summarize each metric across circuit sizes while emphasizing effects achieved by small circuits. Across tasks and models of increasing size, JuntaLearner achieves higher mean CRS compared to attribution baselines on all metrics. Since its cost does not grow with the number of candidate components, JuntaLearner scales to large models while accounting for set-level interactions and avoiding first-order approximations.

We introduce the first learned articulated body simulator based on a selective state space model (SSM), called PhysMamba. PhysMamba predicts next-frame full-body state from position, rotation, and joint-action history, without velocity inputs. We compare four architectures under partial- and full-observation inputs and three training protocols. The from-scratch rollout training protocol gives Mamba2 strong short- and mid-horizon accuracy under partial observation (s10 = 43 mm, 2/50 diverged), while the two-stage teacher-based rollout protocol stabilizes GRU but fails for Mamba2. With CUDA graph compilation, Mamba2 reaches 0.107 ms per frame (9,334 FPS) on an H100 GPU, within 1.1$\times$ of GRU's un-compiled throughput, adding under 1% latency to a 30 Hz HMR pipeline and enabling integration as a differentiable physics module for video-based mesh recovery.

Micro aerial vehicles (MAVs) enable rapid indoor 3D reconstruction for inspection and time-critical situational awareness, but must operate under strict flight-time budgets and safety constraints that require explicit return-to-home (RTH) feasibility with a conservative margin. We present a topology-first active reconstruction framework that decouples high-fidelity rendering from navigation. A dense 3D Gaussian Splatting (3DGS) map is optimized as the reconstruction target, while a lightweight Truncated Signed Distance Field (TSDF)/occupancy scaffold supports conservative collision checking and online construction of a sparse 3D Voronoi skeleton roadmap. Since directly extracting roadmaps from TSDF geometry can be unstable under noisy and incomplete online fusion, we validate nodes and edges using carved free-space consistency and dense visibility checks, which suppress behind-wall phantom structure and stabilize planning. Viewpoints are selected on the roadmap using a flight-time-budget-aware objective and a lightweight receding-horizon lookahead to improve non-myopic exploration behavior. We evaluate on photorealistic indoor simulation benchmarks (ReplicaCAD, Gibson, and HM3D), reporting reconstruction coverage/error, RTH success, and computational cost. Results demonstrate improved or competitive performance relative to recent GS-based baselines under identical flight-time budgets.

Language-model systems can separate contextual computation, persistent storage, and exact execution instead of updating all capabilities through one shared parameter system. We investigate FEM-ASM, a finite-element-method-inspired organization in which independently constructed document states and deterministic executable skills contribute typed proposals to a shared language-model state. An explicit residual operator reconciles proposals attached to common interface nodes. We evaluate this organization through controlled experiments and negative results rather than claiming a physical finite-element formulation of language. An attention-free Multi-Mesh prototype learns causal language modeling but does not establish competitive general capability. A versioned store contains 52,809 reconstructive memory elements near a 1.7-billion-floating-value budget; reconstruction is incomplete, with approximately 75% token accuracy. Support-aware lexical indices make these elements addressable under provenance-controlled query construction. For executable arithmetic, positional result observations substantially improve neural rendering relative to a repeated global result vector, and output substitutions change the model's preferred answer. A bounded attachment demonstration further measures the effect of making selected evidence available, without establishing the utility of loading an entire multi-billion-value store. The results support a separation of storage, execution, and neural coordination, while identifying unresolved limitations in question-only retrieval, unrestricted answer generation, and end-to-end efficiency.

Reinforcement learning with verifiable rewards improves reasoning, while the allocation of learning signal shapes which solutions remain accessible under repeated sampling. Group-relative objectives assign equal advantages to equally rewarded responses, making aggregate credit proportional to sampled mode frequency. We introduce Exploration-Preserving Policy Optimization (ExPPO), a lightweight advantage-shaping rule that redistributes credit using prompt-relative, length-normalized response surprisal and prompt pass rate. ExPPO combines bounded shaping with shared normalization to preserve verifier polarity and approximately maintain each prompt group's total absolute sequence-advantage mass. Our analysis characterizes response-level credit allocation alongside sampled mode updates, deriving local conditions for gains in entropy and correct-mode discovery. Experiments show improved in-domain and out-of-domain reasoning coverage, higher aggregate response accuracy, and strong coverage at large sampling budgets. A controlled multi-answer evaluation further demonstrates increased correct-mode yield and gains in diversity among verified-correct responses. Code is available at https://github.com/jinhangzhan/ExPPO

Zero-sum Markov games arise in a wide variety of sequential decision-making problems such as adversarial learning and planning against modeled uncertainties. However, prior work on finite-sample guarantees on the learned state-action value function ($Q$-function) for zero-sum Markov games is largely restricted to finite-action settings, or to continuous-action games in which agents have linear dynamics and quadratic rewards (i.e. linear-quadratic, or LQ, games). A central challenge in extending such guarantees to more general continuous-action zero-sum Markov games is that the associated Bellman operator involves a minimax problem that need not admit a tractable saddle-point solution. To this end, we first introduce a class of neural network function approximators for the $Q$-function that is convex-concave in the players' actions, which guarantees that this minimax problem admits a pure-strategy saddle point. We then study an online variant of fitted $Q$-iteration employing this function class and establish, to the best of our knowledge, the first finite-sample guarantees for non-LQ zero-sum Markov games with continuous states and actions. Our code can be found at https://github.com/CLeARoboticsLab/QCanPlayThatGame.