Ziwen Xu, Kewei Xu, Haoming Xu, Haiwen Hong, Longtao Huang, Hui Xue, Ningyu Zhang, Yongliang Shen, Guozhou Zheng, Huajun Chen, Shumin Deng. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Generative Retrieval (GR) offers a promising paradigm for recommendation through next-token prediction (NTP). However, scaling it to large-scale industrial systems introduces three challenges: (i) within a single request, the identical model inputs may produce inconsistent outputs due to the pagination request mechanism; (ii) the prohibitive cost of encoding long user behavior sequences with multi-token item representations based on semantic IDs, and (iii) aligning the generative policy with nuanced user preference signals. We present GenRec, a preference-oriented generative framework deployed on the JD App https://www.jd.com that addresses above challenges within a single decoder-only architecture. For training objective, we propose Page-wise NTP task, which supervises over an entire interaction page rather than each interacted item individually, providing denser gradient signal and resolving the one-to-many ambiguity of point-wise training. On the prefilling side, an asymmetric linear Token Merger compresses multi-token Semantic IDs in the prompt while preserving full-resolution decoding, reducing input length by ~2× with negligible accuracy loss. To further align outputs with user satisfaction, we introduce GRPO-SR, a reinforcement learning method that pairs Group Relative Policy Optimization with NLL regularization for training stability, and employs Hybrid Rewards combining a dense reward model with a relevance gate to mitigate reward hacking. In month-long online A/B tests serving production traffic, GenRec achieves 9.5% improvement in click count and 8.7% in transaction count over the existing pipeline.

Abstract We introduce a discrete polydisperse micrograin BSDF model for the rendering of porous surface materials composed of microscopic elements of different size, shape and reflectance distributed on a bulk medium. Our approach generalizes the anisotropic monodisperse micrograin model We first reformulate it in a non‐axis‐aligned configuration, allowing for the later combination of different micrograin types elongated in arbitrary directions. We then extend the monodisperse model to the polydisperse case, deriving its three key components: (i) a general filling factor that controls the mix between micrograins and the bulk medium; (ii) an exact normal distribution function for surfaces composed of polydisperse micrograin distributions; and (iii) the corresponding fully‐correlated shadowing and masking term. This results in an analytical single‐scattering BSDF for discrete polydisperse surface materials, validated over ground truth simulations, for which we also derive a dedicated importance sampling procedure. Our model supports varying heights and anisotropy orientations of different micrograin types as input, giving additional control to simulate phenomena like retro‐reflection from mixed materials, color mixture depending on lighting and observation directions, multiple directions of anisotropy, etc.

Lesion morphology and quantity evaluation in computer tomography (CT) images are critical for precise disease diagnosis. Most existing methods employ machine learning-based methods to separately evaluate the morphology and quantity of individual lesion, neglecting the synergy between morphological structure and quantitative distribution. This limitation presents challenges when handling multiple complex lesions. This paper proposes an unsupervised correlation learning-based clustering model for evaluating lesion morphology and quantity in scenarios involving multiple complex lesions without predefined specific-logic. Specifically, the model utilizes clinical knowledge and changes in the in- or out-degree of lesion regions to learn their interdependencies, automatically recognizing domain-specific morphological features. These morphological features serve as key representations for morphology estimation and provide essential contextual information for quantity analysis. Furthermore, the model perceives quantity evaluation as a density-based clustering process. By interacting with domain-specific morphological features, the model dynamically adjusts the search objects, followed by designing morphology-specific parameter search strategies to autonomously learn spatial relationships between lesion regions. This approach facilitates the exploration of optimal parameters for accurate lesion evaluation without manual intervention. Experiments conducted on the kidney stone dataset including 53 samples and the kidney tumor dataset comprising 300 samples, indicate that the proposed model has achieved 92.45% and 95.33% accuracy in morphology analysis, respectively. For quantity analysis, the proposed model has achieved 79.25% and 94.33% accuracy, outperforming the well-performing AR-DBSCAN method by +30.19% and DRL-DBSCAN method by +6%. The proposed model is demonstrated to be effective in handling morphology and quantity estimation for multiple complex lesions.

The stone recognition and analysis in CT images are significant for automatic kidney stone diagnosis. Although certain contributions have been made, existing methods overlook the promoting effect of clinical knowledge on model performance and clinical interpretation. Thus, it is attractive to establish methods for detecting and evaluating kidney stones originating from the practical diagnostic process. Inspired by this, a novel clinical-inspired framework is proposed to involve the diagnostic process of urologists for better analysis. The diagnostic process contains three main steps, the localization step, the identification step and the evaluation step. Three modules integrating the decision-making mode of urologists are designed to mimic the diagnosis process. The object attention module simulates the localization step to provide the position of kidneys by embedding weight feature factor and angle loss. The feature-driven discriminative module mimics the identification step to detect stones by extracting geometric and positional features. The analysis module based on the principle of clustering and graphic combination is a quantitative analysis strategy for simulating the evaluation step. This work constructed a clinical dataset collecting 27,885 transverse CT images with stones and/or clinical interference. Experiments on the dataset show that the object attention module outperforms the well-performing Yolov7 model by 1% , and the analysis module outperforms the well-performing AR-DBSCAN model and the formula method by 21.9% average cluster accuracy and 17.35% average error. Experiments demonstrate that the proposed framework is recently the most effective solution for recognizing and evaluating kidney stones.

The last industrial revolutions and the digital transformation have led to a rise of robotics and to the emergence of the concept of digital twin. A major challenge falls within the update of this virtual representation, so that the supervision operator and the system itself can take appropriate decisions. One way to achieve that is to take advantage of the multi-robot perception capabilities by merging their individual observations to collectively enhance object recognition and robot environmental understanding. Since object recognition strongly depends on the viewing angles, one challenge deals with identifying the most relevant camera poses containing the most relevant information about the nature of the object. In this paper we propose a smart view selection approach which aims at determining the poses of the cameras and the number of the most informative views while maximising the object recognition. Based on a synthetic view dataset of traditional industrial objects, we adopt a clustering-based approach for maximising the inter-class distance and minimising the intra-class one. To do so, we compute a score for each view based on the Fowlkes-Mallows Index. This leads us to order the dataset and select a subset of views maximising the score. Then, this subset is used as a training dataset for a knn-classifier. The results, presented in terms of F1-score metric, are promising and highlight the relevance of our work: i) our smart selection enables the collection of a limited number of the most informative camera poses for object recognition; ii) feature extraction from a pre-trained CNN combined with a clustering algorithm allows the separability of industrial object categories; iii) our approach is robust since it provides good performances while the camera poses are in the neighbourhood of the exact camera positions provided by our processing pipeline.