arXiv subjects

Kaixuan Yang

Publications and source records attributed to Kaixuan Yang.

6 recordsLinked to original sources

arXiv2026

Location-based Training with Complementary Folded Linear Orderings for Multichannel Speech Separation

Location-based training (LBT) effectively resolves the output permutation problem in multichannel speech separation by imposing deterministic spatial orderings. For planar microphone arrays, LBT typically adopts circular azimuth ordering to cover the full spatial range. However, the resulting cyclic topology introduces a discontinuity at the wrap-around point, increasing learning complexity and limiting the effective use of spatial cues. This work investigates this limitation by introducing location-based training with folded linear orderings (LBT-FLOs), which collapse circular azimuths into controlled linear orderings. While individual LBT-FLOs exhibit front-back ambiguity, each provides enhanced spatial discriminability over specific azimuth regions. Exploiting their complementarity, we propose an ensemble-style framework that selects among multiple LBT-FLOs using azimuth-guided scoring. Experiments across planar array geometries and reverberant conditions show modest but consistent improvements over circular-ordering LBT, with robustness to azimuth estimation errors.

eess.AS↗

arXiv2026

Kwai Keye-VL-2.0 Technical Report

We introduce Kwai Keye-VL-2.0-30B-A3B, an open-source Mixture-of-Experts (MoE) multimodal foundation model designed to advance long-video understanding and agentic intelligence. To address the challenges of ultra-long contexts, information redundancy, and prohibitive computational costs inherent in hour-level videos, Keye-VL-2.0 is the first to adapt DeepSeek Sparse Attention (DSA) to GQA-based multimodal architectures, enabling lossless 256K context processing while capturing critical frames and long-range temporal dependencies. This architecture is underpinned by a highly optimized training and inference infrastructure, including scalable video I/O, heterogeneous ViT-LM parallelism, and custom DSA kernels that significantly maximize throughput and minimize computational overhead. Furthermore, to overcome the algorithmic dilemma of catastrophic forgetting during multi-task alignment, we introduce Cross-Modal Multi-Teacher On-Policy Distillation (MOPD) paired with Context-RL and Video-RL. By distilling dense token-level teacher feedback from on-policy rollouts back into the MoE backbone, which activates only 3B parameters, Keye-VL-2.0 natively empowers advanced agent collaboration across Code, Tool, and Search scenarios with multimodal self-correction. Extensive evaluations across video understanding, temporal grounding, reasoning, STEM, and agent benchmarks demonstrate that Keye-VL-2.0-30B-A3B achieves state-of-the-art performance among models of similar scale, particularly excelling in fine-grained temporal localization on TimeLens and long-video comprehension on Video-MME-v2 and LongVideoBench. We release our model checkpoints to accelerate community progress toward scalable and robust multimodal agentic applications.

cs.CV↗

arXiv2026

Dynamically Slimmable Speech Enhancement Network with Metric-Guided Training

To further reduce the complexity of lightweight speech enhancement models, we introduce a gating-based Dynamically Slimmable Network (DSN). The DSN comprises static and dynamic components. For architecture-independent applicability, we introduce distinct dynamic structures targeting the commonly used components, namely, grouped recurrent neural network units, multi-head attention, convolutional, and fully connected layers. A policy module adaptively governs the use of dynamic parts at a frame-wise resolution according to the input signal quality, controlling computational load. We further propose Metric-Guided Training (MGT) to explicitly guide the policy module in assessing input speech quality. Experimental results demonstrate that the DSN achieves comparable enhancement performance in instrumental metrics to the state-of-the-art lightweight baseline, while using only 73% of its computational load on average. Evaluations of dynamic component usage ratios indicate that the MGT-DSN can appropriately allocate network resources according to the severity of input signal distortion.

eess.AS↗

arXiv2026

ElectriQ: A Benchmark for Assessing the Response Capability of Large Language Models in Power Marketing

As power systems decarbonise and digitalise, high penetrations of distributed energy resources and flexible tariffs make electric power marketing (EPM) a key interface between regulation, system operation and sustainable-energy deployment. Many utilities still rely on human agents and rule- or intent-based chatbots with fragmented knowledge bases that struggle with long, cross-scenario dialogues and fall short of requirements for compliant, verifiable and DR-ready interactions. Meanwhile, frontier large language models (LLMs) show strong conversational ability but are evaluated on generic benchmarks that underweight sector-specific terminology, regulatory reasoning and multi-turn process stability. To address this gap, we present ElectriQ, a large-scale benchmark and evaluation framework for LLMs in EPM. ElectriQ contains over 550k dialogues across six service domains and 24 sub-scenarios and defines a unified protocol that combines human ratings, automatic metrics and two compliance stress tests-Statutory Citation Correctness and Long-Dialogue Consistency. Building on ElectriQ, we propose SEEK-RAG, a retrieval-augmented method that injects policy and domain knowledge during finetuning and inference. Experiments on 13 LLMs show that domain-aligned 7B models with SEEK-RAG match or surpass much larger models while reducing computational cost, providing an auditable, regulation-aware basis for deploying LLM-based EPM assistants that support demand-side management, renewable integration and resilient grid operation.

cs.CL↗

arXiv2025

Direction-aware multi-scale gradient loss for infrared and visible image fusion

Infrared and visible image fusion aims to integrate complementary information from co-registered source images to produce a single, informative result. Most learning-based approaches train with a combination of structural similarity loss, intensity reconstruction loss, and a gradient-magnitude term. However, collapsing gradients to their magnitude removes directional information, yielding ambiguous supervision and suboptimal edge fidelity. We introduce a direction-aware, multi-scale gradient loss that supervises horizontal and vertical components separately and preserves their sign across scales. This axis-wise, sign-preserving objective provides clear directional guidance at both fine and coarse resolutions, promoting sharper, better-aligned edges and richer texture preservation without changing model architectures or training protocols. Experiments on open-source model and multiple public benchmarks demonstrate effectiveness of our approach.

cs.CV↗

arXiv2025

LumiCRS: Asymmetric Contrastive Prototype Learning for Long-Tail Conversational Recommender Systems

Conversational recommender systems (CRSs) often suffer from an extreme long-tail distribution of dialogue data, causing a strong bias toward head-frequency blockbusters that sacrifices diversity and exacerbates the cold-start problem. An empirical analysis of DCRS and statistics on the REDIAL corpus show that only 10% of head movies account for nearly half of all mentions, whereas about 70% of tail movies receive merely 26% of the attention. This imbalance gives rise to three critical challenges: head over-fitting, body representation drift, and tail sparsity. To address these issues, we propose LumiCRS, an end-to-end framework that mitigates long-tail imbalance through three mutually reinforcing layers: (i) an Adaptive Comprehensive Focal Loss (ACFL) that dynamically adjusts class weights and focusing factors to curb head over-fitting and reduce popularity bias; (ii) Prototype Learning for Long-Tail Recommendation, which selects semantic, affective, and contextual prototypes to guide clustering and stabilize body and tail representations; and (iii) a GPT-4o-driven prototype-guided dialogue augmentation module that automatically generates diverse long-tail conversational snippets to alleviate tail sparsity and distribution shift. Together, these strategies enable LumiCRS to markedly improve recommendation accuracy, diversity, and fairness: on the REDIAL and INSPIRED benchmarks, LumiCRS boosts Recall@10 and Tail-Recall@10 by 7-15% over fifteen strong baselines, while human evaluations confirm superior fluency, informativeness, and long-tail relevance. These results demonstrate the effectiveness of multi-layer collaboration in building an efficient and fair long-tail conversational recommender.

cs.AI↗