Each one is read from its source and summarised: what it is, the problem it tackles, and what you could use it for.
Paper2026-09-29
AECSF is a training-free adaptive ensemble conditional score filter designed for Bayesian state estimation in high-dimensional nonlinear systems.
ProblemExisting training-free score filters often rely on heuristic likelihood corrections that compromise posterior accuracy by neglecting uncertainty about the system state associated with each noisy reverse particle.
Use it forHigh-dimensional nonlinear data assimilation; Bayesian state estimation with limited forecast ensembles; Non-Gaussian posterior sampling in dynamical systems
data assimilationscore-based diffusionBayesian estimationnonlinear filteringhigh-dimensional systems
arxiv.org ↗
Paper2026-09-29
This paper introduces Hedge-Cover, an information-theoretic algorithm for smoothed online prediction that achieves sublinear regret in the presence of bounded adversarial responses.
ProblemExisting algorithms for smoothed online regression lacked minimax optimal adaptive regret guarantees when responses are adversarial, leaving an open problem in the theoretical understanding of this framework.
Use it forOnline learning systems where data distribution shifts adversarially but remains smooth relative to a base measure; Theoretical analysis of regret bounds in non-i.i.d. settings; Designing robust prediction algorithms for financial or time-series data with adversarial noise
online-learningregret-boundsadversarial-mltheoretical-cssmoothed-analysis
arxiv.org ↗
Paper2026-09-29
This paper proposes a method for multiobjective Bayesian optimization (MOBO) that uses a multitask Gaussian process to jointly model objectives and constraints.
ProblemExisting MOBO approaches typically evaluate all objectives and constraints in a coupled fashion, ignoring inherent correlations that could allow for more efficient, decoupled evaluation strategies.
Use it forOptimizing complex engineering systems where evaluating all objectives simultaneously is costly or impossible; Hyperparameter tuning for machine learning models with multiple conflicting metrics; Drug discovery or chemical synthesis where different properties must be optimized with limited experimental budget
bayesian-optimizationmultiobjective-optimizatgaussian-processmachine-learningoptimization-theory
arxiv.org ↗
Paper2026-09-29
This paper derives exact finite-width formulas for the expected number of activation switches and scalar kinks in ReLU networks using conditional Kac-Rice theory.
ProblemLack of precise analytical tools to quantify how the piecewise-linear geometry of ReLU networks evolves during training.
Use it forAnalyzing the geometric complexity of neural network decision boundaries; Predicting the number of affine regions in a trained ReLU network; Understanding how supervised learning affects network piecewise-linear structure
neural-networksrelukac-rice-formulaaffine-geometrypiecewise-linear
arxiv.org ↗
Paper2026-09-29
This paper analyzes distributionally robust reinforcement learning (DR-RL) in the average-reward setting under weak communication.
ProblemLack of finite-sample guarantees for distributionally robust reinforcement learning in average-reward settings under weak communication, particularly for different types of uncertainty sets.
Use it forDesigning robust control policies for stochastic systems with model uncertainty; Analyzing sample complexity in average-reward reinforcement learning; Developing algorithms for MDPs with weak communication structures
reinforcement-learningrobust-optimizationaverage-rewardfinite-sample-analysismarkov-decision-processe
arxiv.org ↗
Paper2026-09-29
This paper introduces a doubly-anchored Distributionally Robust Optimization (DRO) framework for domain adaptation that uses the intersection of source and target divergence balls to define its ambiguity set.
ProblemStandard domain adaptation lacks worst-case guarantees, and existing DRO methods ignore available target structure by centering ambiguity sets solely on the source law.
Use it forDomain adaptation tasks where target data is available but potentially corrupted or shifted; Regression problems requiring worst-case risk guarantees under distribution shift; Constructing robust estimators that leverage both source and target covariate structures
domain adaptationdistributionally robust machine learning theorygeneralization boundsregression
arxiv.org ↗
Paper2026-09-29
SCOUT is a method for attributing the source of model distillation using only the current text outputs of a student model.
ProblemExisting distillation attribution methods fail when the student model undergoes further training (SFT, RLHF) after distillation, or when auditors lack access to the pre-distillation checkpoint.
Use it forAuditing the provenance of publicly released LLMs to identify their distillation sources; Detecting unauthorized distillation from commercial APIs; Tracing the persistence of teacher-specific syntactic signatures through post-training stages
model-distillationattributionprovenancellm-safetysyntactic-analysis
arxiv.org ↗
Library2026-09-28
TensorFold is a Python library and CLI tool that serves text models on Apple Silicon (MLX) and NVIDIA GPUs via an OpenAI-compatible API.
ProblemStandard speculative decoding implementations often produce non-deterministic or approximate outputs, and local LLM serving on Apple Silicon lacks efficient, exact decoding pipelines for modern model families.
Use it forServing LLMs locally on Apple Silicon Macs with OpenAI API compatibility; Running speculative decoding for models like Qwen3.8, Nemotron, and GLM-5.3; Deploying quantized (4-bit/EXL3) models with exact arithmetic guarantees
llm-inferenceapple-siliconmlxspeculative-decodingopenai-compatible-api
github.com ↗
Eval/benchmark2026-09-28
RecToolBench is an MCP-based benchmark designed to evaluate agentic recommender systems that must resolve fuzzy user intent using external tools.
ProblemExisting benchmarks assume explicit user intent and isolated function calls, failing to capture the complexity of realistic tool orchestration required when user instructions are fuzzy or ambiguous.
Use it forEvaluating the robustness of LLM agents in handling ambiguous user queries for recommendation tasks; Benchmarking the ability of models to ground semantic parameters in tool calls; Assessing multi-step evidence integration and final recommendation quality in agentic pipelines
recommender-systemsmcpbenchmarkagentic-aitool-orchestration
arxiv.org ↗