Product retrieval is the backbone of e-commerce search: for each user query, it identifies a high-recall candidate set from billions of items, laying the foundation for high-quality ranking and user experience. Despite extensive optimization for mainstream queries, existing systems still struggle with long-tail queries, especially knowledge-intensive ones. These queries exhibit diverse linguistic patterns, often lack explicit purchase intent, and require domain-specific knowledge reasoning for accurate interpretation. They also suffer from a shortage of reliable behavioral logs, which makes such queries a persistent challenge for retrieval optimization.

Abstract Inverse rendering aims to infer scene parameters from observed images. In Solar Power Tower (SPT) systems, this corresponds to an aiming optimization problem—adjusting heliostats' orientations to shape the radiative flux density distribution (RFDD) on the receiver to conform to a desired distribution. The SPT system is widely favored in the field of renewable energy, where aiming optimization is crucial for ensuring its thermal efficiency and safety. However, traditional aiming optimization methods are inefficient and fail to meet online demands. In this paper, a novel optimization approach, DiffNEG, is proposed. DiffNEG introduces a differentiable rasterization method to model the reflected radiative flux of each heliostat as an elliptical Gaussian distribution. It leverages data‐driven techniques to enhance simulation accuracy and employs automatic differentiation combined with gradient descent to achieve online, gradient‐guided optimization in a continuous solution space. Experiments on a real large‐scale heliostat field with nearly 30,000 heliostats demonstrate that DiffNEG can optimize within 10 seconds, improving efficiency by one order of magnitude compared to the latest DiffMCRT method and by three orders of magnitude compared to traditional heuristic methods, while also exhibiting superior robustness under both steady and transient state.

Unsupervised domain adaptation (UDA) aims to alleviate the domain shift by transferring knowledge learned from a labeled source dataset to an unlabeled target domain. Although UDA has seen promising progress recently, it requires access to data from both domains, making it problematic in source data-absent scenarios. In this article, we investigate a practical task source-free domain adaptation (SFDA) that alleviates the limitations of the widely studied UDA in simultaneously acquiring source and target data. In addition, we further study the imbalanced SFDA (ISFDA) problem, which addresses the intra-domain class imbalance and inter-domain label shift in SFDA. We observe two key issues in SFDA that: 1) target data form clusters in the representation space regardless of whether the target data points are aligned with the source classifier and 2) target samples with higher classification confidence are more reliable and have less variation in their classification confidence during adaptation. Motivated by these observations, we propose a unified method, named intrinsic consistency preservation with adaptively reliable samples (ICPR), to jointly cope with SFDA and ISFDA. Specifically, ICPR first encourages the intrinsic consistency in the predictions of neighbors for unlabeled samples with weak augmentation (standard flip-and-shift), regardless of their reliability. ICPR then generates strongly augmented views specifically for adaptively selected reliable samples and is trained to fix the intrinsic consistency between weakly and strongly augmented views of the same image concerning predictions of neighbors and their own. Additionally, we propose to use a prototype-like classifier to avoid the classification confusion caused by severe intra-domain class imbalance and inter-domain label shift. We demonstrate the effectiveness and general applicability of ICPR on six benchmarks of both SFDA and ISFDA tasks. The reproducible code of our proposed ICPR method is available at https://github.com/CFM-MSG/Code_ICPR.