CORTEXA
← Browse
arxivcs.LG2026-06-26

COCOLogic-V2: Identifying Logical Inconsistencies via Truly Hard-Negatives

David Steinmann, Antonia Wüst, Kristian Kersting, Wolfgang Stammer

While interpretable models such as concept bottleneck models (CBMs) and program synthesis methods enable verification of model decisions, their evaluation is typically limited to simple tasks, leaving complex reasoning on real-world images largely unexplored. We introduce COCOLogic-V2, an object-centric dataset for visual inductive reasoning on real-world images covering a broad subset of first-order logic. By categorizing samples into positive variants, near-boundary (NB), and far-from-boundary (FB) negatives, COCOLogic-V2 enables fine-grained diagnosis of model accountability. Our evaluations show that models tend to separate positive and FB samples well but fail on NB samples, while perceptual noise and large rule-induced search spaces pose additional challenges in few-shot settings. Together, these results highlight that visual inductive reasoning remains an open challenge and COCOLogic-V2 provides a concrete foundation for advancing methods in this direction.

View free PDFSource page

Related papers

arxivcs.LGmath.STstat.ML2026-07-24

Beyond Negative-Ridge Endpoints: Mixed-Sign Spectral Regularization via Negative-Shifted Gradient Descent

Peng Zhao

In overparameterized linear regression, many weak spectral directions act like a ridge penalty on the signal-bearing spectrum; negative ridge is the natural correction, pushing filters above one. The stable negative-ridge endpoint, however, is structurally limited: its pole must…

View free PDFSource page
arxivcs.LGstat.ML2026-06-26

Disentangling Continuous-Time Latent Dynamics: Identifiability of Latent SDEs via Diffusion Shifts

Yuanyuan Wang, Wenjie Wang, Haoxuan Li, Mingming Gong, Kun Zhang

Causal representation learning for time series has developed strong identifiability results in discrete-time latent causal models, but identifiability in continuous-time latent stochastic differential equation (SDE) models remains largely open. We address this gap using environme…

View free PDFSource page
arxivcs.LGphysics.geo-ph2026-07-03

MABLE: Masked Autoencoding with Bi-Lipschitz Decoding for Embeddings and Graph Metric Learning

Yaniv Shulman, Shaghayegh Akbarpour, Jack B. Muir

We propose MABLE (Masked Autoencoding with Bi-Lipschitz Decoding for Embeddings and Graph Metric Learning), a self-supervised framework for learning node and graph embeddings from large, heterogeneous graphs, demonstrated here on geospatial mineral-exploration data. MABLE combine…

View free PDFSource page
arxivcs.AIcs.LGcs.SC2026-07-23

Explaining Weather Bulletins via ILP

Enrico Santi, Alessandro Dal Palù, Agostino Dovier, Talissa Dreossi, Andrea Formisano

Inductive Logic Programming (ILP) originated within the Logic Programming community in the Nineties as a framework for combining symbolic learning with declarative knowledge representation. Nowadays, mature ILP frameworks exist and they are capable of learning complex, non-monoto…

View free PDFSource page
arxivcs.ROcs.CLcs.LG2026-06-29

ViTL: Temporal Logic-Guided Zero-Shot Natural Language Navigation via Vision-Language Models

Kaier Liang, Hengde Dai, Cristian-Ioan Vasile

Enabling robots to follow natural language commands to complete zero-shot long-horizon tasks remains challenging. It requires extracting implicit temporal and logical constraints from natural language commands and executing multiple sub-tasks accordingly. Recent zero-shot object…

View free PDFSource page
arxivcs.CLcs.LG2026-07-08

Hallucination Self-Play: Bootstrapping Reinforced Detector via Evolved Generator

Shiping Yang, Shining Liang, Weihao Liu, Wenbiao Ding, Linjun Shou, Lu Cheng, et al.

Identifying faithfulness hallucinations in LLM-generated outputs remains challenging due to the scarcity of high-quality annotated data. Recent work relies on advanced LLMs to synthesize training data, including rationales, labels, and hallucinated claims. However, these methods…

View free PDFSource page