Keyu Li, Junhao Shi, Yang Xiao, Mohan Jiang, Jie Sun, Yunze Wu, Dayuan Fu, Shijie Xia, Xiaojie Cai, Tianze Xu, Weiye Si, Wenjie Li, Dequan Wang, Pengfei Liu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Aligning powerful AI models on tasks that surpass human evaluation capabilities is the central problem of superalignment. To address this problem, weak-to-strong generalization aims to elicit the capabilities of strong models through weak supervisors and ensure that the behavior of strong models aligns with the intentions of weak supervisors without unsafe behaviors such as deception. Although weak-to-strong generalization exhibiting certain generalization capabilities, strong models exhibit significant overfitting in weak-to-strong generalization: Due to the strong fit ability of strong models, erroneous labels from weak supervisors may lead to overfitting in strong models. In addition, simply filtering out incorrect labels may lead to a degeneration in question quality, resulting in a weak generalization ability of strong models on hard questions. To mitigate overfitting in weak-to-strong generalization, we propose a two-stage framework that simultaneously improves the quality of supervision signals and the quality of input questions. Experimental results in three series of large language models and two mathematical benchmarks demonstrate that our framework significantly improves PGR (Performance Gap Recovered) compared to naive weak-to-strong generalization, even achieving up to 100% PGR on some models.

Large Vision-Language Models (LVLMs) show promise for embodied planning tasks but struggle with complex scenarios involving unfamiliar environments and multi-step goals. Current approaches rely on environment-agnostic imitation learning that disconnects instructions from environmental contexts, causing models to struggle with context-sensitive instructions and rely on supplementary cues rather than visual reasoning during long-horizon interactions. In this work, we propose World-Aware Planning Narrative Enhancement (WAP), a framework that infuses LVLMs with comprehensive environmental understanding through four cognitive capabilities (visual appearance modeling, spatial reasoning, functional abstraction, and syntactic grounding) while developing and evaluating models using only raw visual observations through curriculum learning. Evaluations on the EB-ALFRED benchmark demonstrate substantial improvements, with Qwen2.5-VL achieving a 60.7 absolute improvement in task success rates—particularly in commonsense reasoning (+60.0) and long-horizon planning (+70.0). Notably, our enhanced open-source models outperform proprietary systems like GPT-4o and Claude-3.5-Sonnet by a large margin.

Online action detection aims to identify ongoing actions within untrimmed video streams, with extensive applications in real-life scenarios. However, in practical applications, video frames are received sequentially over time and new action categories continually emerge, giving rise to the challenge of catastrophic forgetting - a problem that remains inadequately explored. Generally, in the field of video understanding, researchers address catastrophic forgetting through class-incremental learning. Nevertheless, online action detection is based solely on historical observations, thus demanding higher temporal modeling capabilities for class-incremental learning methods. In this paper, we conceptualize this task as Class-Incremental Online Action Detection (CIOAD) and propose a novel framework, TS-ILM, to address it. Specifically, TS-ILM consists of two components: task-level temporal pattern extractor and temporal-sensitive exemplar selector. The former extracts the temporal patterns of actions in different tasks and saves them, allowing the data to be comprehensively observed on a temporal level before it is input into the backbone. The latter selects a set of frames with the highest causal relevance and minimum information redundancy for subsequent replay, enabling the model to learn the temporal information of previous tasks more effectively. We benchmark our approach against SoTA class-incremental learning methods applied in the image and video domains on THUMOS'14 and TVSeries datasets. Our method outperforms the previous approaches.

Locally repairable codes (LRC) have dominated the coding design of erasure coding in wide stripes due to their excellent local property and low bandwidth requirement. In distributed storage systems, repair degree usually refers to the number of helper nodes that need to be contacted in data repair process. As the stripe becomes wider, the probability of multiple nodes failing simultaneously in the LRC increases, which leads to a significant increase in the repair degree. So how to improve the existing schemes to improve code reliability and reduce the repair degree in case of multiple failures remains to be further explored. In this paper, we present a new family of locally repairable codes, TFR-LRC (Locally Repairable Codes for seeking the Trade-off between Fault tolerance and Repair degree), which can achieve different trade-offs between fault tolerance and repair cost. TFR-LRC can sacrifice a little bit of storage cost for lower repair degree, and can sacrifice a little bit of repair cost for higher fault tolerance. In order to verify the feasibility of TFR-LRC, we conducted simulation experiments. The results show that, compared to the existing LRC schemes, TFR-LRC can greatly reduce the repair degree in case of multiple failures by sacrificing a little bit of storage cost without changing the traditional LRC structures, and meanwhile TFR-LRC can make higher fault tolerance under the same parameter settings by sacrificing a little bit of repair cost.

We present a vision-based gesture interaction technique that is capable of tracking both hands in real-time. By only recovering features calculated from images, large amount of computation for 3D reconstructing and features tracking in 3D voxel grid is avoided. Thus, our system becomes an ideal tool for bimanual 3D gesture interaction, which needs neither markers nor gloves. Experimental results show that our method can complete the same pipeline with much less computational resource compared to existing ones. What's more, the application interface shows the effectiveness and intuitiveness of our bimanual tracking method.

In this paper we study the testability of circuits derived from Binary Decision Diagrams (BDDs) under the bridging fault model. It is shown that testability can be formulated in terms of symbolic BDD operations. By this, test pattern generation can be carried out in polynomial time. A technique to improve testability is presented. Experimental results show that a complete classification can be carried out very efficiently.

The complexity of integrated circuits is rapidly growing. This leads to more and more time and money spent on the test of these circuits. Besides minimizing the logic needed for a given function the testability of the resulting circuit becomes a major issue during synthesis. One way to synthesize a circuit for a given function is to directly convert the binary decision diagram (BDD) of that function into a circuit. It is known that optimizations of the BDD transfer to the derived circuit. Therefore in this paper we evaluate different optimization techniques for BDDs based on variable reordering with respect to the path delay fault testability of the resulting circuit. We show an optimization strategy that allows to compromise during synthesis between logic size and testability.

We present a technique to derive fully testable circuits under the stuck-at fault model (SAFM) and the path-delay fault model (PDFM). Starting from a function description as a binary decision diagram, the netlist is generated by a linear time mapping algorithm. Only one additional input and one inverter are needed to achieve 100% testable circuits under SAFM and PDFM. Experiments are given to show the advantages of the technique.

A new technique for synthesizing totally symmetric Boolean functions is presented that achieves complete robust path delay fault testability. BDDs are used for the synthesis step. Only one additional input and one inverter are needed to achieve 100% Path Delay Fault (PDF) testability. The size of the circuit is guaranteed to be at most quadratic in the number of inputs. The test vectors for any PDF can be generated in linear time. Experimental results underline the efficiency of the technique. In contrast to previous approaches, the technique can also be applied to multi-output functions.

We present a technique to derive fully testable circuits under the Stuck-At Fault Model (SAFM) and the Path-Delay Fault Model (PDFM). Starting from a function description as a Binary Decision Diagram (BDD) the netlist is generated by a linear time mapping algorithm. Only one additional input and one inverter are needed to achieve 100% testable circuits under SAFM and PDFM. Experiments are given to show the advantages of the the technique in comparison to previously presented methods.