Kaixuan Yang

Kaixuan Yang, Wei Xiang, Zhenshuai Chen et al.

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.

Jinzhi Wang, Qingke Peng, Haozhou Li et al.

Electric power marketing customer service plays a critical role in addressing inquiries, complaints, and service requests. However, current systems, such as China's 95598 hotline, often struggle with slow response times, inflexible procedures, and limited accuracy in domain-specific tasks. While large language models (LLMs) like GPT-4o and Claude 3 demonstrate strong general capabilities, they lack the domain expertise and empathy required in this field. To bridge this gap, we introduce ElectriQ, the first benchmark designed to evaluate and enhance LLMs in electric power marketing scenarios. ElectriQ consists of a dialogue dataset covering six key service categories and introduces four evaluation metrics: professionalism, popularity, readability, and user-friendliness. We further incorporate a domain-specific knowledge base and propose a knowledge augmentation method to boost model performance. Experiments on 13 LLMs reveal that smaller models such as LLama3-8B, when fine-tuned and augmented, can surpass GPT-4o in terms of professionalism and user-friendliness. ElectriQ establishes a comprehensive foundation for developing LLMs tailored to the needs of power marketing services.

Jinzhi Wang, Bin Li, Qingke Peng et al.

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.