The paper introduces a formal definition of the structural role of latent factors in dynamical systems, describing where a latent factor is expressed, whether it receives influence from past observations, and whether it transmits influence to future observations. It addresses the challenge of correspondence across different structural identifications and proves that, under certain conditions, all three roles can be jointly identified using a common permutation at each time point. Building on this theory, the authors present RoleTrace, a framework that traces these structural roles in a shared latent coordinate system, and validate its effectiveness through synthetic experiments and real-world multiparty meeting data.
arXiv:2608.28991v1 Announce Type: cross Abstract: Causal representation learning (CRL) aims to recover latent causal variables and their structural relations from high-dimensional observations. Exist...
The paper introduces BRAID, a hierarchical latent-variable model that generates multi-person human motion by explicitly modeling both group-level interaction dynamics and individual behavior conditioned on evolving group context. It treats social motion generation as a meta-transfer learning problem, learning shared interaction priors across datasets and adapting them to arbitrary context sets of observed people and joints. BRAID supports coherent generation under full, sparse, or partial observations and produces compact social-state vectors useful for downstream embodied-agent systems, with evaluations on social forecasting, tracking, in-filling, and response generation.
arXiv:2609.36852v1 Announce Type: new Abstract: Trajectory prediction is a key component for understanding human behavior patterns in dynamic scenes. Researchers have devoted substantial efforts to m...
The paper investigates how large language model (LLM) agents converge toward agreement during multi‑turn conversations about subjective topics. By training a linear classifier on conversation activations, the authors uncover an internal representation of agreement that varies across turns and does not reliably match external agreement at the individual conversation level. Aggregating this internal agreement across conversations reveals a Markovian structure: external agreement at each turn depends directly on the previous turn, and internal agreement couples to external agreement via a power‑law mapping, with models often expressing external agreement before their internal representations shift.
As LLM agents interact in the wild and shape each other's decisions, understanding how they converge toward agreement becomes important. We study this in multi-turn conversations where two LLMs discus...
arXiv:2605. 15995v2 Announce Type: replace-cross Abstract: Learning latent representations from complex data is central to modern machine learning, spanning temporal, multimodal, and partially observed systems.
The paper studies when joint-embedding predictive architectures (JEPAs) can recover underlying causal states from high‑dimensional observations. It introduces a latent variable model where observations arise from causal states with action‑conditioned dynamics, and proposes an information‑theoretic objective that maximizes conditional likelihood while preserving state entropy. The authors prove identifiability conditions—particularly sufficient action‑induced variation—and instantiate the objective as an action‑modulated Gaussian additive‑noise model (A‑JEPA), demonstrating theoretical and empirical success in synthetic and visual benchmarks.
arXiv:2609.35955v1 Announce Type: new Abstract: Understanding human-entity interactions requires recovering each person-action event's participants, roles, and shared identities. This structure can s...
arXiv:2608. 16245v1 Announce Type: new Abstract: Disentangled representation learning seeks latent representations whose indicidual dimensions each align with a distinct covariate.
arXiv:2608. 06549v1 Announce Type: cross Abstract: LLMs are increasingly being applied to tasks involving institutional and political texts, but existing benchmarks evaluate them on isolated documents or single tasks.
MOSAIC is a sparse temporal variational autoencoder that merges causal representation learning with support recovery to identify interpretable latent modules in scientific time series. It leverages regime‑conditioned temporal variation to discover latent variables and uses an additive decoder to recover a sparse set of observations linked to each latent, providing module‑level interpretability. The method offers theoretical guarantees for ANOVA main‑effect identifiability under smooth mixing functions and demonstrates empirical success across diverse domains such as RNA dynamics, solar wind, ENSO climate, industrial process control, and synthetic tokamak data.
The paper investigates how large language models encode and use relational information among tokens across transformer layers. By analyzing activations from prompts that require inferring relationships among three cyclic tokens (months, hours, weekdays, musical notes), the authors find a consistent layerwise progression: intermediate layers capture pairwise relationships, while later layers encode the full three‑token relationship to predict the next token. They also identify geometrically structured token relationships that do not influence prediction, and show that constraining models to use only causally relevant joint representations improves next‑token accuracy.