analiza aiqbench
This paper introduces CoCoEvolve, a self-supervised framework designed to align chart images, tabular data, and visualization code by enforcing explicit one-to-one correspondences rather than treating them as ambiguous many-to-many mappings. By optimizing for consistency across these modalities during both training and inference without requiring additional annotations, the method addresses fundamental challenges in cross-representation understanding where supervision is typically costly or unavailable. The approach demonstrates improved performance on multiple benchmarks, offering a principled signal that enhances generalizability beyond task-specific objectives.
Consistency-Driven Co-Evolution for Self-Supervised Cross-Representation Learning
arXiv:2608.04926v1 Announce Type: cross Abstract: As chart images, tabular data, and visualization code play increasingly important roles across diverse domains, cross-representation understanding across these modalities poses fundamental challenges for AI systems: the relationships across representations are inherently \textit{one-to-many}, supervision is ambiguous and costly, and model optimization lacks a principled signal that is both direction-adaptive and representation-generalizable beyond task-specific objectives. We introduce CoCoEvolve to improve consistency across chart, table, and code representations. Instead of treating cross-representation mapping as a one-to-many problem, we define explicit one-to-one correspondences and optimize models using agreement between representations, without additional annotations. During training, CoCoEvolve@Train performs co-evolution across the chart-table-code cycle, while CoCoEvolve@Test applies the same consistency objective at inference time for test-time co-optimization. We also present CoCoEvolve@Eval, an evaluation suite covering all six cross-representation tasks. Across four benchmarks, CoCoEvolve improves performance in both training-time and test-time settings. Our project page: https://xhguo7.github.io/CoCoEvolve/.
Abstract:As chart images, tabular data, and visualization code play increasingly important roles across diverse domains, cross-representation understanding across these modalities poses fundamental challenges for AI systems: the relationships across representations are inherently \textit{one-to-many}, supervision is ambiguous and costly, and model optimization lacks a principled signal that is both direction-adaptive and representation-generalizable beyond task-specific objectives. We introduce CoCoEvolve to improve consistency across chart, table, and code representations. Instead of treating cross-representation mapping as a one-to-many problem, we define explicit one-to-one correspondences and optimize models using agreement between representations, without additional annotations. During training, CoCoEvolve@Train performs co-evolution across the chart-table-code cycle, while CoCoEvolve@Test applies the same consistency objective at inference time for test-time co-optimization. We also present CoCoEvolve@Eval, an evaluation suite covering all six cross-representation tasks. Across four benchmarks, CoCoEvolve improves performance in both training-time and test-time settings. Our project page: this https URL.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
arXiv:2608.04926 [cs.LG]
(or arXiv:2608.04926v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2608.04926 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 243 cuvinte
Subiecte conexe
D3O: Dynamic Distribution Distillation for Ordinal Regression
arXiv · 28 iul. 2026
RECTOR: Masked Region-Channel-Temporal Modeling for Affective and Cognitive Representation Learning
arXiv · 16 iun. 2026
CORD: Learning Reusable Degradation Representations Across Heterogeneous Physical Systems
arXiv · 1 oct. 2026
Models Can't Remember Their Training. Neither Can You.
LessWrong · 19 iul. 2026