Zhixuan Zhang

Yukun Zheng, Jiang Bian, Guanghao Meng et al.

In large-scale e-commerce platforms like Taobao, it is a big challenge to retrieve products that satisfy users from billions of candidates. This has been a common concern of academia and industry. Recently, plenty of works in this domain have achieved significant improvements by enhancing embedding-based retrieval (EBR) methods, including the Multi-Grained Deep Semantic Product Retrieval (MGDSPR) model [16] in Taobao search engine. However, we find that MGDSPR still has problems of poor relevance and weak personalization compared to other retrieval methods in our online system, such as lexical matching and collaborative filtering. These problems promote us to further strengthen the capabilities of our EBR model in both relevance estimation and personalized retrieval. In this paper, we propose a novel Multi-Objective Personalized Product Retrieval (MOPPR) model with four hierarchical optimization objectives: relevance, exposure, click and purchase. We construct entire-space multi-positive samples to train MOPPR, rather than the single-positive samples for existing EBR models.We adopt a modified softmax loss for optimizing multiple objectives. Results of extensive offline and online experiments show that MOPPR outperforms the baseline MGDSPR on evaluation metrics of relevance estimation and personalized retrieval. MOPPR achieves 0.96% transaction and 1.29% GMV improvements in a 28-day online A/B test. Since the Double-11 shopping festival of 2021, MOPPR has been fully deployed in mobile Taobao search, replacing the previous MGDSPR. Finally, we discuss several advanced topics of our deeper explorations on multi-objective retrieval and ranking to contribute to the community.

Fen Wang, Zekai Shao, Qiman Kang et al.

Chart descriptions are essential for accessibility, cross-modal retrieval, and assisting readers in extracting insights from complex visualizations. As multimodal large language models (MLLMs) are increasingly adopted for automated chart description generation, a critical question arises: how faithfully and insightfully do these models actually describe charts? Current benchmarks fall short on two fronts: existing datasets consist of simple, homogeneous charts paired with shallow, fact-enumerating descriptions; and prevailing metrics fail to capture the multi-faceted nature of description quality. To address these gaps, we present the Chart Faithfulness and Insightfulness Benchmark (ChartFI-Bench). We first summarize four dimensions that characterize high-quality chart descriptions: factual accuracy, salient feature emphasis, domain-informed guidance, and chart-text complementarity. Guided by these dimensions, we construct a high-quality benchmark comprising 896 chart-description pairs, which feature visually complex charts and semantically rich descriptions. Furthermore, we design four aligned evaluation metrics -- Faithfulness, Coverage, Informativeness, and Acuity -- to systematically assess the quality of descriptions across these dimensions. Experiments conducted on mainstream MLLMs demonstrate the effectiveness of the proposed framework and reveal common weaknesses among existing models.

Xingyue Wang, Bo Liu, Meng Wang et al.

Visual Question Answering (VQA) holds great promise for clinical support, particularly in ophthalmology, where retinal fundus photography is essential for diagnosis. However, ophthalmic VQA benchmarks primarily emphasize answer accuracy, neglecting the explicit visual evidence necessary for clinical interpretability. In this work, we introduce FundusGround, a new benchmark for clinically interpretable ophthalmic VQA with spatially-grounded lesion evidence. Specifically, we propose a three-stage pipeline that collects 10,719 fundus images with 15,595 image-level meticulously annotated lesions. To ensure anatomical consistency and clinical validity, all lesions are spatially localized using the Early Treatment Diabetic Retinopathy Study (ETDRS) grid, enabling standardized mapping to nine clinically meaningful retinal regions. Built upon this structured lesion evidence, 72,706 questions are then generated spanning four formats: open-ended, closed-ended, single-choice, and multiple-choice. We further benchmark multiple general- and medical- large vision-language models using dual metrics for answer accuracy and lesion-level reasoning. The experiments demonstrate that incorporating lesion-level visual evidence consistently improves model performance and transparency, highlighting the necessity of explicit spatial grounding for reliable and explainable ophthalmic VQA.

Bin Wang, Fei Deng, Shuang Wang et al.

Semantic segmentation of remote sensing (RS) images is a challenging yet essential task with broad applications. While deep learning, particularly supervised learning with large-scale labeled datasets, has significantly advanced this field, the acquisition of high-quality labeled data remains costly and time-intensive. Unsupervised domain adaptation (UDA) provides a promising alternative by enabling models to learn from unlabeled target domain data while leveraging labeled source domain data. Recent self-training (ST) approaches employing pseudo-label generation have shown potential in mitigating domain discrepancies. However, the application of ST to RS image segmentation remains underexplored. Factors such as variations in ground sampling distance, imaging equipment, and geographic diversity exacerbate domain shifts, limiting model performance across domains. In that case, existing ST methods, due to significant domain shifts in cross-domain RS images, often underperform. To address these challenges, we propose integrating contrastive learning into UDA, enhancing the model's ability to capture semantic information in the target domain by maximizing the similarity between augmented views of the same image. This additional supervision improves the model's representational capacity and segmentation performance in the target domain. Extensive experiments conducted on RS datasets, including Potsdam, Vaihingen, and LoveDA, demonstrate that our method, SimSeg, outperforms existing approaches, achieving state-of-the-art results. Visualization and quantitative analyses further validate SimSeg's superior ability to learn from the target domain. The code is publicly available at https://github.com/woldier/SiamSeg.

Liren Yu, Caiyuan Li, Feiyi Dong et al.

Recent advances in Large Language Models (LLMs) have inspired a surge of scaling law research in industrial search, advertising, and recommendation systems. However, existing approaches focus mainly on architectural improvements, overlooking the critical synergy between data and architecture design. We observe that scaling model parameters alone exhibits diminishing returns, i.e., the marginal gain in performance steadily declines as model size increases, and that the performance degradation caused by complex heterogeneous data distributions is often irrecoverable through model design alone. In this paper, we propose UniScale to address these limitation, a novel co-design framework that jointly optimizes data and architecture to unlock the full potential of model scaling, which includes two core parts: (1) ES$^3$ (Entire-Space Sample System), a high-quality data scaling system that expands the training signal beyond conventional sampling strategies from both intra-domain request contexts with global supervised signal constructed by hierarchical label attribution and cross-domain samples aligning with the essence of user decision under similar content exposure environment in search domain; and (2) HHSFT (Heterogeneous Hierarchical Sample Fusion Transformer), a novel architecture designed to effectively model the complex heterogeneous distribution of scaled data and to harness the entire space user behavior data with Heterogeneous Hierarchical Feature Interaction and Entire Space User Interest Fusion, thereby surpassing the performance ceiling of structure-only model tuning. Extensive experiments on large-scale real world E-commerce search platform demonstrate that UniScale achieves significant improvements through the synergistic co-design of data and architecture and exhibits clear scaling trends, delivering substantial gains in key business metrics.

Zhixuan Zhang, Pingyu Wang, Xingjian Zheng et al.

Transferable attacks generate adversarial examples on surrogate models to fool unknown victim models, posing real-world threats and growing research interest. Despite focusing on flat losses for transferable adversarial examples, recent studies still fall into suboptimal regions, especially the flat-yet-sharp areas, termed as deceptive flatness. In this paper, we introduce a novel black-box gradient-based transferable attack from a perspective of dual-order information. Specifically, we feasibly propose Adversarial Flatness (AF) to the deceptive flatness problem and a theoretical assurance for adversarial transferability. Based on this, using an efficient approximation of our objective, we instantiate our attack as Adversarial Flatness Attack (AFA), addressing the altered gradient sign issue. Additionally, to further improve the attack ability, we devise MonteCarlo Adversarial Sampling (MCAS) by enhancing the inner-loop sampling efficiency. The comprehensive results on ImageNet-compatible dataset demonstrate superiority over six baselines, generating adversarial examples in flatter regions and boosting transferability across model architectures. When tested on input transformation attacks or the Baidu Cloud API, our method outperforms baselines.

Zhixuan Zhang, Chi Zhang, Zhenning Niu et al.

In this manuscript, we introduce a semi-automatic scene graph annotation tool for images, the GeneAnnotator. This software allows human annotators to describe the existing relationships between participators in the visual scene in the form of directed graphs, hence enabling the learning and reasoning on visual relationships, e.g., image captioning, VQA and scene graph generation, etc. The annotations for certain image datasets could either be merged in a single VG150 data-format file to support most existing models for scene graph learning or transformed into a separated annotation file for each single image to build customized datasets. Moreover, GeneAnnotator provides a rule-based relationship recommending algorithm to reduce the heavy annotation workload. With GeneAnnotator, we propose Traffic Genome, a comprehensive scene graph dataset with 1000 diverse traffic images, which in return validates the effectiveness of the proposed software for scene graph annotation. The project source code, with usage examples and sample data is available at https://github.com/Milomilo0320/A-Semi-automatic-Annotation-Software-for-Scene-Graph, under the Apache open-source license.

Zhi Sun, Wenming Zhang, Yi Wei et al.

Large Language Models (LLMs) are equipped with profound semantic knowledge, making them a natural choice for injecting semantic generalization into personalized search systems. However, in practice we find that directly fine-tuning LLMs on industrial personalized tasks (e.g. next item prediction) often yields suboptimal results. We attribute this bottleneck to a critical Knowledge--Action Gap: the inherent conflict between preserving pre-trained semantic knowledge and aligning with specific personalized actions by discriminative objectives. Empirically, action-only training objectives induce Semantic Collapse, such as attention ``sinks''. This degradation severely cripples the LLM's generalization, failing to bring improvements to personalized search systems. We propose KARMA (Knowledge--Action Regularized Multimodal Alignment), a unified framework that treats semantic reconstruction as a train-only regularizer. KARMA optimizes a next-interest embedding for retrieval (Action) while enforcing semantic decodability (Knowledge) through two complementary objectives: (i) history-conditioned semantic generation, which anchors optimization to the LLM's native next-token distribution, and (ii) embedding-conditioned semantic reconstruction, which constrains the interest embedding to remain semantically recoverable. On Taobao search system, KARMA mitigates semantic collapse (attention-sink analysis) and improves both action metrics and semantic fidelity. In ablations, semantic decodability yields up to +22.5 HR@200. With KARMA, we achieve +0.25 CTR AUC in ranking, +1.86 HR in pre-ranking and +2.51 HR in recalling. Deployed online with low inference overhead at ranking stage, KARMA drives +0.5% increase in Item Click.

Bin Wang, Fei Deng, Peifan Jiang et al.

Low-dose computed tomography (LDCT) has become the technology of choice for diagnostic medical imaging, given its lower radiation dose compared to standard CT, despite increasing image noise and potentially affecting diagnostic accuracy. To address this, advanced deep learning-based LDCT denoising algorithms have been developed, primarily using Convolutional Neural Networks (CNNs) or Transformer Networks with the Unet architecture. This architecture enhances image detail by integrating feature maps from the encoder and decoder via skip connections. However, current methods often overlook enhancements to the Unet architecture itself, focusing instead on optimizing encoder and decoder structures. This approach can be problematic due to the significant differences in feature map characteristics between the encoder and decoder, where simple fusion strategies may not effectively reconstruct images.In this paper, we introduce WiTUnet, a novel LDCT image denoising method that utilizes nested, dense skip pathways instead of traditional skip connections to improve feature integration. WiTUnet also incorporates a windowed Transformer structure to process images in smaller, non-overlapping segments, reducing computational load. Additionally, the integration of a Local Image Perception Enhancement (LiPe) module in both the encoder and decoder replaces the standard multi-layer perceptron (MLP) in Transformers, enhancing local feature capture and representation. Through extensive experimental comparisons, WiTUnet has demonstrated superior performance over existing methods in key metrics such as Peak Signal-to-Noise Ratio (PSNR), Structural Similarity (SSIM), and Root Mean Square Error (RMSE), significantly improving noise removal and image quality.

Zhengxing Cheng, Yuheng Huang, Zhixuan Zhang et al.

Given the ubiquity of multi-task in practical systems, Multi-Task Learning (MTL) has found widespread application across diverse domains. In real-world scenarios, these tasks often have different priorities. For instance, In web search, relevance is often prioritized over other metrics, such as click-through rates or user engagement. Existing frameworks pay insufficient attention to the prioritization among different tasks, which typically adjust task-specific loss function weights to differentiate task priorities. However, this approach encounters challenges as the number of tasks grows, leading to exponential increases in hyper-parameter tuning complexity. Furthermore, the simultaneous optimization of multiple objectives can negatively impact the performance of high-priority tasks due to interference from lower-priority tasks. In this paper, we introduce a novel multi-task learning framework employing Lagrangian Differential Multiplier Methods for step-wise multi-task optimization. It is designed to boost the performance of high-priority tasks without interference from other tasks. Its primary advantage lies in its ability to automatically optimize multiple objectives without requiring balancing hyper-parameters for different tasks, thereby eliminating the need for manual tuning. Additionally, we provide theoretical analysis demonstrating that our method ensures optimization guarantees, enhancing the reliability of the process. We demonstrate its effectiveness through experiments on multiple public datasets and its application in Taobao search, a large-scale industrial search ranking system, resulting in significant improvements across various business metrics.

Zhixuan Zhang, Xiaopeng Li, Qi Liu

Background subtraction (BGS) is utilized to detect moving objects in a video and is commonly employed at the onset of object tracking and human recognition processes. Nevertheless, existing BGS techniques utilizing deep learning still encounter challenges with various background noises in videos, including variations in lighting, shifts in camera angles, and disturbances like air turbulence or swaying trees. To address this problem, we design a spiking autoencoder network, termed SAEN-BGS, based on noise resilience and time-sequence sensitivity of spiking neural networks (SNNs) to enhance the separation of foreground and background. To eliminate unnecessary background noise and preserve the important foreground elements, we begin by creating the continuous spiking conv-and-dconv block, which serves as the fundamental building block for the decoder in SAEN-BGS. Moreover, in striving for enhanced energy efficiency, we introduce a novel self-distillation spiking supervised learning method grounded in ANN-to-SNN frameworks, resulting in decreased power consumption. In extensive experiments conducted on CDnet-2014 and DAVIS-2016 datasets, our approach demonstrates superior segmentation performance relative to other baseline methods, even when challenged by complex scenarios with dynamic backgrounds.

Malu Zhang, Jiadong Wang, Burin Amornpaisannon et al.

Spiking Neural Networks (SNNs) use spatio-temporal spike patterns to represent and transmit information, which is not only biologically realistic but also suitable for ultra-low-power event-driven neuromorphic implementation. Motivated by the success of deep learning, the study of Deep Spiking Neural Networks (DeepSNNs) provides promising directions for artificial intelligence applications. However, training of DeepSNNs is not straightforward because the well-studied error back-propagation (BP) algorithm is not directly applicable. In this paper, we first establish an understanding as to why error back-propagation does not work well in DeepSNNs. To address this problem, we propose a simple yet efficient Rectified Linear Postsynaptic Potential function (ReL-PSP) for spiking neurons and propose a Spike-Timing-Dependent Back-Propagation (STDBP) learning algorithm for DeepSNNs. In STDBP algorithm, the timing of individual spikes is used to convey information (temporal coding), and learning (back-propagation) is performed based on spike timing in an event-driven manner. Our experimental results show that the proposed learning algorithm achieves state-of-the-art classification accuracy in single spike time based learning algorithms of DeepSNNs. Furthermore, by utilizing the trained model parameters obtained from the proposed STDBP learning algorithm, we demonstrate the ultra-low-power inference operations on a recently proposed neuromorphic inference accelerator. Experimental results show that the neuromorphic hardware consumes 0.751~mW of the total power consumption and achieves a low latency of 47.71~ms to classify an image from the MNIST dataset. Overall, this work investigates the contribution of spike timing dynamics to information encoding, synaptic plasticity and decision making, providing a new perspective to design of future DeepSNNs and neuromorphic hardware systems.