Wrist PPG heart rate monitoring is vulnerable to motion artifacts, and naively applying corrections can lower average error while damaging individual windows. This paper proposes a selective correction framework organized as an agent around a fixed core estimator: signal-analysis modules extract pulse and motion evidence, a temporal memory module generates candidate heart rates, and a decision module predicts each candidate's benefit and harm before a separate risk gate chooses to retain, correct, or reject the output. The authors train the auxiliary modules on PPG-DaLiA and evaluate frozen weights on three additional public datasets plus a prototype wearable stream, reporting macro-averaged MAE reductions up to 44.5%. The core idea—explicitly trading off predicted gains against harms rather than universally replacing estimates—is a clear contribution to wearable heart rate analysis, though deployment readiness is weakened by calibration transfer failures and an extremely limited prototype.

LLM agents can execute prohibited actions when tool responses omit safety-relevant facts, even if every returned statement is true and the user request is benign. EpiSEM-Bench introduces 300 matched unsafe/safe task pairs across six domains with controlled evidence suppression and an optional audit-based recovery tool. The work shows that robustness to missing evidence and useful recovery are distinct capabilities that single-score benchmarks conceal, making joint evaluation essential for safe agent deployment.

This paper studies handover prediction in cellular networks, where time-series models must meet a hard 40 ms inference deadline and a tight energy budget on user equipment. Rather than ranking by accuracy alone, the authors propose a protocol that filters models by tail latency feasibility then ranks survivors on measured per-inference energy—including feature computation. They benchmark eight architectures and find that gradient-boosted trees dominate the accuracy–energy–latency frontier, while the top accuracy-only neural model (CNN-1D, F1=0.914) costs over 3× more energy and trails in F1. The work challenges standard benchmarks by showing that duty-cycled measurement and full-pipeline accounting can invert model rankings.

This paper shows that large language models fail to treat co-referring molecular identifiers—such as trivial names, IUPAC names, and SMILES strings—as interchangeable routes to the same knowledge. The authors introduce ChemIDBench, a controlled benchmark of 896 approved drugs, and demonstrate a striking double dissociation: lexical names unlock entity-associated facts like therapeutic indications, while structural identifiers unlock structure-computable properties like benzene-ring presence. Even when a model recognizes a molecule from its SMILES, the corresponding factual knowledge often remains inaccessible, a phenomenon the authors term cross-identifier binding failure. They trace this to a missing access relation rather than missing knowledge, and propose Binding Self-Distillation to transfer lexical access directly to SMILES without retraining from scratch.

xVGAE tackles graph generation for moderate-sized, potentially dense graphs where diffusion and autoregressive methods become prohibitively slow. It builds on the Aldous-Hoover representation theorem to design a hierarchical VAE with a global graph latent $z_0$ and node latents $z_i$ that are conditionally i.i.d., using a learnable neural graphon decoder for one-shot generation. The core pitch is that this structure yields sample quality competitive with state-of-the-art while sampling one to three orders of magnitude faster, and produces interpretable latent representations.

Standard on-policy distillation forces a student to copy a teacher at every token, even when the teacher's preferred actions hurt the student's eventual success. This paper proposes Advantage-Based On-Policy Distillation (AOPD), which weights each teacher token by its counterfactual advantage under the current student policy. The key insight is that imitation should optimize for student task reward, not teacher agreement, and the authors derive an unbiased estimator that needs only standard student rollouts and terminal rewards.

Onger and Foscari analyze competitive market making as a two-sided pricing game where strategic liquidity providers face a non-strategic informed trader with inelastic demand. Because bid and ask quotes jointly set the market mid-price that determines trade direction, standard equilibrium existence theorems fail and payoffs are non-convex and discontinuous. The authors prove that pure-strategy Nash equilibria exist via a novel fixed-point argument on a midpoint operator, derive a structural linear scaling of spreads with asset volatility, and show that no-internal-regret learning on fine price grids recovers the competitive benchmark as tick size vanishes.

Depth growth and looping reuse computation across layers to train or run Transformers more efficiently. This paper shows that both methods create an inductive bias toward learning from context over memorizing fixed associations in weights, accelerating in-context learning and propagating induction heads through depth. The evidence spans models up to 3B parameters and connects architecture design to the mechanistic origins of few-shot performance.

The paper tests whether matching spectral predictability $\Omega$ between synthetic pretraining data and target time series is sufficient for effective transfer learning under distribution shift. They evaluate this "predictability-aware" hypothesis using LSTM, PatchTST, and TCN across controlled synthetic conditions and the real-world ETTm1 benchmark. The core finding is a negative result: while $\Omega$-matching works under ideal conditions, it fails across different synthetic source families on real data, suggesting practitioners need more nuanced source-selection criteria.

Standard deep Gaussian processes become increasingly ill-conditioned as depth grows because discrete layer composition amplifies parameter perturbations. This paper proposes replacing the discrete stack with a latent continuous-depth SDE whose drift family carries a GP-based prior, yielding two models: CDGP with a direct separable GP prior, and FlowDGP with a flow-evolved base GP. The central claim is that depth in probabilistic regression is better modeled as continuous latent evolution than as a discrete composition, and that retaining a GP prior over the drift improves stability and competitiveness relative to both discrete DGPs and neural continuous-depth alternatives.

Gradual domain adaptation (GDA) tackles large distribution shifts by splitting them into ordered intermediate steps, but existing methods often suffer from error accumulation and unstable adversarial training. This paper proposes FlowGDA, a flow-matching framework that models gradual adaptation as continuous transport on a statistical manifold in a representation space, using time-dependent velocity fields learned via a simulation-free regression objective. By coupling local left-to-right transport with a source-anchored semantic consistency loss, the method aims to avoid the error accumulation and adversarial instability of discrete stepwise GDA methods.

Representation learning has improved sample efficiency in unconstrained MDPs, but extending it to constrained settings with unknown dynamics remains largely open. This paper studies low-rank constrained MDPs with continuous states where latent transition features must be learned online while controlling cumulative constraint violations. The authors propose two primal-dual algorithms that combine maximum-likelihood feature learning, uncertainty bonuses, and decaying $\epsilon$-greedy auxiliary probing to obtain the first sublinear regret and violation guarantees for this setting, with an explicit tradeoff between return-policy optimality and the safety cost of exploratory execution.

This paper adapts the Feature Selection Layer (FSL) into a post-hoc interpretability tool for pre-trained neural networks on tabular data. Instead of training an embedded feature weighting layer jointly with the network, the authors freeze the pre-trained model and insert a lightweight diagonal scaling layer that learns per-feature weights via fine-tuning. The approach aims to let practitioners retrofit already deployed models to obtain feature importance scores without retraining from scratch, bridging embedded feature selection and post-hoc attribution methods.

RiboBridge addresses the challenge of designing RNA molecules that bind to target protein structures when neither the native RNA sequence/structure nor the binding interface is known. Its core idea is to separate RNA conformation generation in a pretrained discrete structural space (RiboSphere) from protein-relative assembly via predicted cross-frame interaction geometry and rigid placement. This decomposition matters because high-quality protein–RNA complex data are scarce and existing methods entangle conformational sampling with interface discovery.

CausalCoT addresses the problem of unfaithful chain-of-thought reasoning in LLMs, where dense attention allows models to bypass their own generated rationales and exploit global context instead. It enforces procedural causal dependence by decomposing queries into a directed acyclic graph of sub-questions and evaluating each node in an isolated context containing only its direct parents. The framework promises more faithful explanations and precise test-time steerability via local interventions.

DIAL tackles the challenge of fine-grained control over instruction adherence—how strongly an LLM should follow instructions like "Keep it short." The core idea is to distill classifier-free guidance (CFG) into a lightweight, text-conditioned steering adapter that edits hidden states at a single layer. This matters because it promises scalable, efficient, adjustable adherence across diverse instructions without requiring labeled responses or per-instruction fitting.

Iterative combinatorial auctions require many costly rounds to discover clearing prices. This paper proposes FastCombo, an offline deep RL method that learns a recurrent graph policy to post sequential item prices with a single neural-network forward pass at deployment. The core insight is that maximizing immediate clearing probability can be arbitrarily suboptimal in expected rounds, so formulating the task as a POMDP and training over complete auction trajectories lets the policy acquire information that speeds up later clearing. Results show competitive clearing on standard CATS benchmarks with 300–900× lower online computation than a Bayesian baseline, though consistent round savings remain elusive outside a controlled one-item domain.

Instrumental convergence predicts that capable agents develop power-seeking subgoals, but it is unclear whether reinforcement learning with verifiable rewards (RLVR) monotonically increases such behavior in language models. This paper evaluates OLMo-3-7B-Think across sixteen post-training checkpoints using InstrumentalEval and finds a nonmonotonic trajectory: instrumental-like outputs rise during early RLVR, peak near step 200, and then fall below the SFT baseline by the final checkpoint. The result challenges simple "more RL equals more convergence" narratives and shows that checkpoint-level monitoring can reveal dynamics hidden by endpoint comparisons.

AgenticEvolve replaces the one-shot LLM variation operator in evolutionary program search with a persistent agent that edits code, runs self-designed experiments, and commits candidates for formal evaluation. By letting the agent test hypotheses locally between expensive evaluations, the framework aims to improve sample efficiency and final program quality on tasks ranging from online bin packing to symbolic regression. The authors report gains over a matched No-analysis ablation and over OpenEvolve, while noting that cost trade-offs depend heavily on model pricing.

This paper challenges how digital twins for type 1 diabetes are evaluated. Instead of the standard practice of scoring models by glucose trajectory error (RMSE), the authors propose measuring decision quality: whether a twin's simulated trajectories correctly rank candidate insulin treatments. They benchmark six methods on 30 virtual patients and show that a model can predict glucose accurately yet rank treatments no better than random, while a population-average baseline that ignores individual data ranks surprisingly well. The result is a convincing argument that clinical decision-support models need decision-aware evaluation.