Knowledge catalogue
arXiv:2510.22767v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) typically come with a fixed architecture, despite growing evidence that not all layers contribute equally to ever
arXiv:2507.11198v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) enable new possibilities for qualitative research at scale, including annotation and qualitative coding of educat
arXiv:2605.08153v1 Announce Type: new Abstract: With the rapid development of machine learning applications on time-series data, accurately assessing the value of training samples has become essential
arXiv:2605.10388v1 Announce Type: new Abstract: End to end (E2E) autonomous driving trajectory prediction is often trained with camera frames sampled at the highest available temporal frequency, assum
arXiv:2512.13618v3 Announce Type: replace Abstract: Representing continuous time is a critical and under-explored challenge in modeling temporal event sequences with large language models (LLMs). Vari
arXiv:2605.09967v1 Announce Type: new Abstract: While researchers are finding concepts represented as linear directions in language models, a bag of linear directions fails to capture relational struc
arXiv:2605.09046v1 Announce Type: new Abstract: In many real-world robotic tasks, robots must generate dynamically feasible motions that reliably reach desired goals even under uncertainty. Yet existi
arXiv:2605.09329v1 Announce Type: new Abstract: Speculative decoding accelerates LLM inference by using a fast draft model to generate tokens and a more accurate target model to verify them. Its perfo
arXiv:2605.08215v1 Announce Type: new Abstract: Visual Foresight VLA (VF-VLA) has become a prominent architectural choice in the recent VLA due to its impressive performance. Nevertheless, the inheren
arXiv:2605.08181v1 Announce Type: cross Abstract: Parameter-efficient fine-tuning methods introduce a small number of training parameters, enabling pre-trained models to adapt rapidly to new data dist
arXiv:2601.02366v2 Announce Type: replace-cross Abstract: Graph-based recommendation has achieved great success in recent years. The classical graph recommendation model utilizes ID embedding to store
arXiv:2605.06047v2 Announce Type: replace-cross Abstract: Tabular foundation models (TFMs), such as TabPFN-2.6, TabICLv2, ConTextTab, Mitra, LimiX, and TabDPT, achieve strong zero-shot performance thr
arXiv:2505.11577v4 Announce Type: replace-cross Abstract: Recent application programming interface (API) restrictions on major social media platforms challenge compliance with the EU Digital Services
arXiv:2605.10754v1 Announce Type: new Abstract: LLM-based foundation agents that perceive, reason, and act across thousands of reasoning steps are rapidly becoming the dominant paradigm for deploying
arXiv:2605.10334v1 Announce Type: new Abstract: Recent deepfake detection methods demonstrate improved cross-dataset generalization, yet the underlying mechanisms remain underexplored. We introduce th
arXiv:2605.09225v1 Announce Type: cross Abstract: Jailbreak attacks -- adversarial prompts that bypass LLM alignment through purely linguistic manipulation -- pose a growing operational security threa
arXiv:2605.09838v1 Announce Type: new Abstract: Sentiment analysis has been of long-standing interest in psychotherapy research. Recently, the Transformer deep learning architecture has produced text-
arXiv:2602.23928v2 Announce Type: replace Abstract: We show that large language models (LLMs) have an astonishing ability to recover meaning from severely degraded English texts. Texts in which conten
arXiv:2605.08427v1 Announce Type: new Abstract: Self-play red team is an established approach to improving AI safety in which different instances of the same model play attacker and defender roles in
arXiv:2605.05775v2 Announce Type: replace-cross Abstract: We report the design and results of the third autoPET challenge (MICCAI 2024), which benchmarked automated lesion segmentation in whole-body P
arXiv:2605.10237v1 Announce Type: new Abstract: We study how temporal correlations in the data can make certain sparse learning problems efficiently learnable by gradient-based methods. Our focus is o
arXiv:2605.10698v1 Announce Type: cross Abstract: Multi-agent systems (MAS) assume that collaborating inherently improves Large Language Model (LLM) reasoning. We challenge this by demonstrating that
arXiv:2605.08666v1 Announce Type: new Abstract: A commonly accepted explanation of critic-free RL for LLMs, based on sequence-level rewards, is that it reinforces successful rollouts with a positive a
arXiv:2605.09883v1 Announce Type: cross Abstract: As current Multimodal Large Language Models rapidly saturate canonical visual reasoning benchmarks, a key question emerges: do these strong scores gen
arXiv:2502.01237v3 Announce Type: replace Abstract: Direct Alignment Algorithms (DAAs) simplify LLM alignment by directly optimizing policies, bypassing reward modeling and RL. While DAAs differ in th
arXiv:2605.09020v1 Announce Type: new Abstract: This paper presents a novel Direct Integration Theorem (DIT), derived as a non-trivial corollary of the classical Central Slice Theorem (CST). The DIT p
arXiv:2603.05653v2 Announce Type: replace-cross Abstract: Adolescents spend an increasing amount of their time in digital environments where their still-developing cognitive capacities leave them unab
arXiv:2605.08611v1 Announce Type: new Abstract: Current language model memory systems store what happened but not how it felt. This distinction -- between semantic memory (knowing about a past event)
arXiv:2512.18928v3 Announce Type: replace Abstract: This work introduces a novel nonlinear optimal filtering method, termed the Ensemble Schr{odinger Bridge nonlinear filter. The proposed filter combi
arXiv:2605.08737v1 Announce Type: cross Abstract: On-policy distillation (OPD) is widely used for LLM post-training. When pushed with a reward-extrapolation coefficient lambda > 1, the student can lif
arXiv:2605.10687v1 Announce Type: new Abstract: Turbulent dynamical systems are characterized by nonlinear interactions and stochastic effects that generate coupled statistical quantities, such as non
arXiv:2605.10828v1 Announce Type: new Abstract: As large language models are increasingly deployed in retrieval-augmented generation and agentic systems that accumulate extensive context, understandin
arXiv:2605.10851v1 Announce Type: new Abstract: We introduce the Generalized Turing Test (GTT), a formal framework for comparing the capabilities of arbitrary agents via indistinguishability. For agen
arXiv:2601.18832v3 Announce Type: replace-cross Abstract: Scaling test-time compute enhances long chain-of-thought (CoT) reasoning, yet existing approaches face a fundamental trade-off between computa
arXiv:2605.08464v1 Announce Type: new Abstract: The manifold hypothesis (MH) is often used to explain how machine learning can overcome the curse of dimensionality. However, the MH is only applicable
arXiv:2605.09887v1 Announce Type: cross Abstract: Sparse autoencoders (SAEs) operationalise the linear representation hypothesis: they reconstruct model activations as sparse linear combinations of in
arXiv:2605.09195v1 Announce Type: new Abstract: Large language models confidently produce outdated answers, and no existing method can detect them. We show this is not an engineering failure but a str
arXiv:2605.08746v1 Announce Type: new Abstract: In training a neural network with gradient descent (GD), each iteration induces a linear operator that governs first-order updates to a model's internal
arXiv:2605.09900v1 Announce Type: new Abstract: A vision-language model can look at a knot diagram and report what it sees, yet fail to act on that structure. KnotBench pairs an 858,318-image corpus f
arXiv:2605.08837v1 Announce Type: cross Abstract: Abstract concepts - justice, theory, availability - have no single perceivable referent; in the human brain, their meaning emerges from a web of exper
arXiv:2605.10216v1 Announce Type: new Abstract: Native Language Identification (NLI) is the task of determining an author's native language (L1) from their non-native writings. With the advent of huma
arXiv:2510.15995v3 Announce Type: replace-cross Abstract: We study overpricing in a repeated game between two representative agents: a market maker, who controls market liquidity, and a market taker,
arXiv:2605.10799v1 Announce Type: cross Abstract: Corruption studies, the primary tool for evaluating chain-of-thought (CoT) faithfulness, identify which chain positions are 'computationally important
arXiv:2605.09844v1 Announce Type: new Abstract: The Metacognitive Probe is an exploratory five-task, 15-slot diagnostic that decomposes an LLM's confidence behaviour into five behaviourally-distinct d
arXiv:1706.00476v4 Announce Type: replace-cross Abstract: In this paper, we propose a low-rank coordinate descent approach to structured semidefinite programming with diagonal constraints. The approac
arXiv:2605.09916v1 Announce Type: cross Abstract: We introduce the observable Wasserstein distance, a framework for deriving lower bounds on the Wasserstein distance between probability measures on Po
arXiv:2605.10601v1 Announce Type: new Abstract: AI deployment in sensitive domains such as health care, credit, employment, and criminal justice is often treated as unsafe to authorize until model int
arXiv:2605.09221v1 Announce Type: cross Abstract: Fairness impossibility results often look like distinct scalar incompatibility statements. We show that several share one RKHS geometry: fairness crit
arXiv:2605.10393v1 Announce Type: new Abstract: The counting power of Message Passing Neural Networks (MPNN) has been the subject of many recent papers, showing that they can express logic that involv
arXiv:2605.08390v1 Announce Type: new Abstract: Sequence prediction methods for dynamical systems with long memory, i.e. marginally stable systems, typically achieve regret that grows polynomially wit
arXiv:2512.18610v3 Announce Type: replace Abstract: Intuitively, a more deterministic time series should be easier to forecast. However, point-wise loss functions (e.g., MSE and MAE), serving as diffe
arXiv:2605.08529v1 Announce Type: new Abstract: Modern deep learning treats neural networks primarily as endpoint functions from inputs to outputs. Inspired by the shift from force to geometry in phys
arXiv:2511.02623v2 Announce Type: replace Abstract: Post-training alignment of large language models (LLMs) relies on large-scale human annotations guided by policy specifications that change over tim
arXiv:2605.08323v1 Announce Type: cross Abstract: Communication is fundamental to sustaining reciprocity and cooperation in strategic interactions. We identify and formulate the influence attribution
arXiv:2605.08116v1 Announce Type: cross Abstract: Recent work on text diffusion models offers a promising alternative to autoregressive generation, but controlling their safety remains underexplored.
arXiv:2602.10868v2 Announce Type: replace Abstract: We study the sample complexity of learning a uniform approximation of an n-dimensional cumulative distribution function (CDF) within an error epsilo
arXiv:2605.09739v1 Announce Type: cross Abstract: Large Language Models are increasingly used as zero-shot classifiers in complex reasoning tasks. However, standard constrained decoding suffers from a
arXiv:2605.09330v1 Announce Type: cross Abstract: Agentic memory enables LLMs to persist information beyond a single context window and reuse it in later decisions, but it also introduces a new vulner
arXiv:2605.09969v1 Announce Type: cross Abstract: How should hidden states generated autoregressively be collapsed into a representation that reflects a language model's internal state? Despite tokens
arXiv:2605.10019v1 Announce Type: cross Abstract: Generative models trained on finite data face a fundamental tension: their score-matching or next-token objective converges to the empirical training