Searcharxiv⌕ Search

arXiv subjects

Xiao Zhou

Publications and source records attributed to Xiao Zhou.

At least 109 records · Page 6Linked to original sources
arXiv2024

Knowledge-enhanced Visual-Language Pretraining for Computational Pathology

In this paper, we consider the problem of visual representation learning for computational pathology, by exploiting large-scale image-text pairs gathered from public resources, along with the domain-specific knowledge in pathology. Specifically, we make the following contributions: (i) We curate a pathology knowledge tree that consists of 50,470 informative attributes for 4,718 diseases requiring pathology diagnosis from 32 human tissues. To our knowledge, this is the first comprehensive structured pathology knowledge base; (ii) We develop a knowledge-enhanced visual-language pretraining approach, where we first project pathology-specific knowledge into latent embedding space via a language model, and use it to guide the visual representation learning; (iii) We conduct thorough experiments to validate the effectiveness of our proposed components, demonstrating significant performance improvement on various downstream tasks, including cross-modal retrieval, zero-shot classification on pathology patches, and zero-shot tumor subtyping on whole slide images (WSIs).

cs.CV↗
arXiv2024

Leveraging Web-Crawled Data for High-Quality Fine-Tuning

Most large language models are fine-tuned using either expensive human-annotated data or GPT-4 generated data which cannot guarantee performance in certain domains. We argue that although the web-crawled data often has formatting errors causing semantic inaccuracies, it can still serve as a valuable source for high-quality supervised fine-tuning in specific domains without relying on advanced models like GPT-4. To this end, we create a paired training dataset automatically by aligning web-crawled data with a smaller set of high-quality data. By training a language model on this dataset, we can convert web data with irregular formats into high-quality ones. Our experiments show that training with the model-transformed data yields better results, surpassing training with only high-quality data by an average score of 9.4% in Chinese math problems. Additionally, our 7B model outperforms several open-source models larger than 32B and surpasses well-known closed-source models such as GPT-3.5, highlighting the efficacy of our approach.

cs.CL↗
arXiv2024

Integrated Intention Prediction and Decision-Making with Spectrum Attention Net and Proximal Policy Optimization

For autonomous driving in highly dynamic environments, it is anticipated to predict the future behaviors of surrounding vehicles (SVs) and make safe and effective decisions. However, modeling the inherent coupling effect between the prediction and decision-making modules has been a long-standing challenge, especially when there is a need to maintain appropriate computational efficiency. To tackle these problems, we propose a novel integrated intention prediction and decision-making approach, which explicitly models the coupling relationship and achieves efficient computation. Specifically, a spectrum attention net is designed to predict the intentions of SVs by capturing the trends of each frequency component over time and their interrelations. Fast computation of the intention prediction module is attained as the predicted intentions are not decoded to trajectories in the executing process. Furthermore, the proximal policy optimization (PPO) algorithm is employed to address the non-stationary problem in the framework through a modest policy update enabled by a clipping mechanism within its objective function. On the basis of these developments, the intention prediction and decision-making modules are integrated through joint learning. Experiments are conducted in representative traffic scenarios, and the results reveal that the proposed integrated framework demonstrates superior performance over several deep reinforcement learning (DRL) baselines in terms of success rate, efficiency, and safety in driving tasks.

cs.RO↗
arXiv2024

CGAP: Urban Region Representation Learning with Coarsened Graph Attention Pooling

The explosion of massive urban data recently has provided us with a valuable opportunity to gain deeper insights into urban regions and the daily lives of residents. Urban region representation learning emerges as a crucial realm for fulfilling this task. Among deep learning approaches, graph neural networks (GNNs) have shown promise, given that city elements can be naturally represented as nodes with various connections between them as edges. However, many existing GNN approaches encounter challenges such as over-smoothing and limitations in capturing information from nodes in other regions, resulting in the loss of crucial urban information and a decline in region representation performance. To address these challenges, we leverage urban graph structure information and introduce a hierarchical graph pooling process called Coarsened Graph Attention Pooling (CGAP). CGAP features local attention units to create coarsened intermediate graphs and global features. Additionally, by incorporating urban region graphs and global features into a global attention layer, we harness relational information to enhance representation effectiveness. Furthermore, CGAP integrates region attributes such as Points of Interest (POIs) and inter-regional contexts like human mobility, enabling the exploitation of multi-modal urban data for more comprehensive representation learning. Experiments on three downstream tasks related to the UN Sustainable Development Goals validate the effectiveness of region representations learned by our approach. Experimental results and analyses demonstrate that CGAP excels in various socioeconomic prediction tasks compared to competitive baselines.

cs.SI↗
arXiv2024

MuseCL: Predicting Urban Socioeconomic Indicators via Multi-Semantic Contrastive Learning

Predicting socioeconomic indicators within urban regions is crucial for fostering inclusivity, resilience, and sustainability in cities and human settlements. While pioneering studies have attempted to leverage multi-modal data for socioeconomic prediction, jointly exploring their underlying semantics remains a significant challenge. To address the gap, this paper introduces a Multi-Semantic Contrastive Learning (MuseCL) framework for fine-grained urban region profiling and socioeconomic prediction. Within this framework, we initiate the process by constructing contrastive sample pairs for street view and remote sensing images, capitalizing on the similarities in human mobility and Point of Interest (POI) distribution to derive semantic features from the visual modality. Additionally, we extract semantic insights from POI texts embedded within these regions, employing a pre-trained text encoder. To merge the acquired visual and textual features, we devise an innovative cross-modality-based attentional fusion module, which leverages a contrastive mechanism for integration. Experimental results across multiple cities and indicators consistently highlight the superiority of MuseCL, demonstrating an average improvement of 10% in $R^2$ compared to various competitive baseline models. The code of this work is publicly available at https://github.com/XixianYong/MuseCL.

cs.CV↗
arXiv2024

Rethinking Overlooked Aspects in Vision-Language Models

Recent advancements in large vision-language models (LVLMs), such as GPT4-V and LLaVA, have been substantial. LLaVA's modular architecture, in particular, offers a blend of simplicity and efficiency. Recent works mainly focus on introducing more pre-training and instruction tuning data to improve model's performance. This paper delves into the often-neglected aspects of data efficiency during pre-training and the selection process for instruction tuning datasets. Our research indicates that merely increasing the size of pre-training data does not guarantee improved performance and may, in fact, lead to its degradation. Furthermore, we have established a pipeline to pinpoint the most efficient instruction tuning (SFT) dataset, implying that not all SFT data utilized in existing studies are necessary. The primary objective of this paper is not to introduce a state-of-the-art model, but rather to serve as a roadmap for future research, aiming to optimize data usage during pre-training and fine-tuning processes to enhance the performance of vision-language models.

cs.CV↗
arXiv2024

Ins-HOI: Instance Aware Human-Object Interactions Recovery

Accurately modeling detailed interactions between human/hand and object is an appealing yet challenging task. Current multi-view capture systems are only capable of reconstructing multiple subjects into a single, unified mesh, which fails to model the states of each instance individually during interactions. To address this, previous methods use template-based representations to track human/hand and object. However, the quality of the reconstructions is limited by the descriptive capabilities of the templates so that these methods are inherently struggle with geometry details, pressing deformations and invisible contact surfaces. In this work, we propose an end-to-end Instance-aware Human-Object Interactions recovery (Ins-HOI) framework by introducing an instance-level occupancy field representation. However, the real-captured data is presented as a holistic mesh, unable to provide instance-level supervision. To address this, we further propose a complementary training strategy that leverages synthetic data to introduce instance-level shape priors, enabling the disentanglement of occupancy fields for different instances. Specifically, synthetic data, created by randomly combining individual scans of humans/hands and objects, guides the network to learn a coarse prior of instances. Meanwhile, real-captured data helps in learning the overall geometry and restricting interpenetration in contact areas. As demonstrated in experiments, our method Ins-HOI supports instance-level reconstruction and provides reasonable and realistic invisible contact surfaces even in cases of extremely close interaction. To facilitate the research of this task, we collect a large-scale, high-fidelity 3D scan dataset, including 5.2k high-quality scans with real-world human-chair and hand-object interactions. The code and data will be public for research purposes.

cs.CV↗
arXiv2024

Learning Weakly Supervised Audio-Visual Violence Detection in Hyperbolic Space

In recent years, the task of weakly supervised audio-visual violence detection has gained considerable attention. The goal of this task is to identify violent segments within multimodal data based on video-level labels. Despite advances in this field, traditional Euclidean neural networks, which have been used in prior research, encounter difficulties in capturing highly discriminative representations due to limitations of the feature space. To overcome this, we propose HyperVD, a novel framework that learns snippet embeddings in hyperbolic space to improve model discrimination. Our framework comprises a detour fusion module for multimodal fusion, effectively alleviating modality inconsistency between audio and visual signals. Additionally, we contribute two branches of fully hyperbolic graph convolutional networks that excavate feature similarities and temporal relationships among snippets in hyperbolic space. By learning snippet representations in this space, the framework effectively learns semantic discrepancies between violent and normal events. Extensive experiments on the XD-Violence benchmark demonstrate that our method outperforms state-of-the-art methods by a sizable margin.

cs.CV↗
arXiv2024

Ada-Retrieval: An Adaptive Multi-Round Retrieval Paradigm for Sequential Recommendations

Retrieval models aim at selecting a small set of item candidates which match the preference of a given user. They play a vital role in large-scale recommender systems since subsequent models such as rankers highly depend on the quality of item candidates. However, most existing retrieval models employ a single-round inference paradigm, which may not adequately capture the dynamic nature of user preferences and stuck in one area in the item space. In this paper, we propose Ada-Retrieval, an adaptive multi-round retrieval paradigm for recommender systems that iteratively refines user representations to better capture potential candidates in the full item space. Ada-Retrieval comprises two key modules: the item representation adapter and the user representation adapter, designed to inject context information into items' and users' representations. The framework maintains a model-agnostic design, allowing seamless integration with various backbone models such as RNNs or Transformers. We perform experiments on three widely used public datasets, incorporating five powerful sequential recommenders as backbone models. Our results demonstrate that Ada-Retrieval significantly enhances the performance of various base models, with consistent improvements observed across different datasets. Our code and data are publicly available at: https://github.com/ll0ruc/Ada-Retrieval.

cs.IR↗
arXiv2024

Causally Aware Generative Adversarial Networks for Light Pollution Control

Artificial light plays an integral role in modern cities, significantly enhancing human productivity and the efficiency of civilization. However, excessive illumination can lead to light pollution, posing non-negligible threats to economic burdens, ecosystems, and human health. Despite its critical importance, the exploration of its causes remains relatively limited within the field of artificial intelligence, leaving an incomplete understanding of the factors contributing to light pollution and sustainable illumination planning distant. To address this gap, we introduce a novel framework named Causally Aware Generative Adversarial Networks (CAGAN). This innovative approach aims to uncover the fundamental drivers of light pollution within cities and offer intelligent solutions for optimal illumination resource allocation in the context of sustainable urban development. We commence by examining light pollution across 33,593 residential areas in seven global metropolises. Our findings reveal substantial influences on light pollution levels from various building types, notably grasslands, commercial centers and residential buildings as significant contributors. These discovered causal relationships are seamlessly integrated into the generative modeling framework, guiding the process of generating light pollution maps for diverse residential areas. Extensive experiments showcase CAGAN's potential to inform and guide the implementation of effective strategies to mitigate light pollution. Our code and data are publicly available at https://github.com/zhangyuuao/Light_Pollution_CAGAN.

cs.CY↗
arXiv2023

On the complexity of list $\mathcal H$-packing for sparse graph classes

The problem of packing as many subgraphs isomorphic to $H \in \mathcal H$ as possible in a graph for a class $\mathcal H$ of graphs is well studied in the literature. Both vertex-disjoint and edge-disjoint versions are known to be NP-complete for $H$ that contains at least three vertices and at least three edges, respectively. In this paper, we consider ``list variants'' of these problems: Given a graph $G$, an integer $k$, and a collection $\mathcal L_{\mathcal H}$ of subgraphs of $G$ isomorphic to some $H \in \mathcal H$, the goal is to compute $k$ subgraphs in $\mathcal L_{\mathcal H}$ that are pairwise vertex- or edge-disjoint. We show several positive and negative results, focusing on classes of sparse graphs, such as bounded-degree graphs, planar graphs, and bounded-treewidth graphs.

cs.DS↗
arXiv2023

Can GPT-4V(ision) Serve Medical Applications? Case Studies on GPT-4V for Multimodal Medical Diagnosis

Driven by the large foundation models, the development of artificial intelligence has witnessed tremendous progress lately, leading to a surge of general interest from the public. In this study, we aim to assess the performance of OpenAI's newest model, GPT-4V(ision), specifically in the realm of multimodal medical diagnosis. Our evaluation encompasses 17 human body systems, including Central Nervous System, Head and Neck, Cardiac, Chest, Hematology, Hepatobiliary, Gastrointestinal, Urogenital, Gynecology, Obstetrics, Breast, Musculoskeletal, Spine, Vascular, Oncology, Trauma, Pediatrics, with images taken from 8 modalities used in daily clinic routine, e.g., X-ray, Computed Tomography (CT), Magnetic Resonance Imaging (MRI), Positron Emission Tomography (PET), Digital Subtraction Angiography (DSA), Mammography, Ultrasound, and Pathology. We probe the GPT-4V's ability on multiple clinical tasks with or without patent history provided, including imaging modality and anatomy recognition, disease diagnosis, report generation, disease localisation. Our observation shows that, while GPT-4V demonstrates proficiency in distinguishing between medical image modalities and anatomy, it faces significant challenges in disease diagnosis and generating comprehensive reports. These findings underscore that while large multimodal models have made significant advancements in computer vision and natural language processing, it remains far from being used to effectively support real-world medical applications and clinical decision-making. All images used in this report can be found in https://github.com/chaoyi-wu/GPT-4V_Medical_Evaluation.

cs.CV↗
arXiv2023

Curriculum Proximal Policy Optimization with Stage-Decaying Clipping for Self-Driving at Unsignalized Intersections

Unsignalized intersections are typically considered as one of the most representative and challenging scenarios for self-driving vehicles. To tackle autonomous driving problems in such scenarios, this paper proposes a curriculum proximal policy optimization (CPPO) framework with stage-decaying clipping. By adjusting the clipping parameter during different stages of training through proximal policy optimization (PPO), the vehicle can first rapidly search for an approximate optimal policy or its neighborhood with a large parameter, and then converges to the optimal policy with a small one. Particularly, the stage-based curriculum learning technology is incorporated into the proposed framework to improve the generalization performance and further accelerate the training process. Moreover, the reward function is specially designed in view of different curriculum settings. A series of comparative experiments are conducted in intersection-crossing scenarios with bi-lane carriageways to verify the effectiveness of the proposed CPPO method. The results show that the proposed approach demonstrates better adaptiveness to different dynamic and complex environments, as well as faster training speed over baseline methods.

cs.RO↗
arXiv2023

The MI-Motion Dataset and Benchmark for 3D Multi-Person Motion Prediction

3D multi-person motion prediction is a challenging task that involves modeling individual behaviors and interactions between people. Despite the emergence of approaches for this task, comparing them is difficult due to the lack of standardized training settings and benchmark datasets. In this paper, we introduce the Multi-Person Interaction Motion (MI-Motion) Dataset, which includes skeleton sequences of multiple individuals collected by motion capture systems and refined and synthesized using a game engine. The dataset contains 167k frames of interacting people's skeleton poses and is categorized into 5 different activity scenes. To facilitate research in multi-person motion prediction, we also provide benchmarks to evaluate the performance of prediction methods in three settings: short-term, long-term, and ultra-long-term prediction. Additionally, we introduce a novel baseline approach that leverages graph and temporal convolutional networks, which has demonstrated competitive results in multi-person motion prediction. We believe that the proposed MI-Motion benchmark dataset and baseline will facilitate future research in this area, ultimately leading to better understanding and modeling of multi-person interactions.

cs.CV↗
arXiv2023

Properties of the optical component stars in X-ray binary systems

We collected a total of 4,058 X-ray binary stars, out of which 339 stars had three atmospheric parameters for optical companions from Gaia and LAMOST, while 264 stars had masses and radii of optical companions determined using stellar evolution models. We conducted a thorough discussion on the reliability of each parameter. The statistical analysis revealed a noticeable bimodal distribution in the mass, radius, and age of the optical components. Our findings led to the proposal of a new quantitative classification criterion for X-ray binary stars. In this classification, one type is categorized as high-mass, high-temperature, and young, while the other type is classified as low-mass, low-temperature, and old, corresponding to High-Mass X-Ray Binaries (HMXBs) and Low-Mass X-Ray Binaries (LMXBs), respectively. The dividing lines were established at 11,500 K, 1.7 $M_{\odot}$, and 0.14 Gyr. The classification results of the three parameters showed consistency with one another in 90\% of the cases and were in agreement with previous classifications in 80\% of the cases. We found that the parameters of the two types had well-defined boundaries and distinct patterns. Based on our findings, we suggest that temperature is the best parameter for classification. Therefore, we propose that X-ray binary stars should be classified into high-temperature X-ray binaries (HTXBs) and low-temperature X-ray binaries (LTXBs). We believe that this classification is more convenient in practice and aligns well with physics.

astro-ph.SR↗
arXiv2023

Reconstructing human activities via coupling mobile phone data with location-based social networks

In the era of big data, the ubiquity of location-aware portable devices provides an unprecedented opportunity to understand inhabitants' behavior and their interactions with the built environments. Among the widely used data resources, mobile phone data is the one passively collected and has the largest coverage in the population. However, mobile operators cannot pinpoint one user within meters, leading to the difficulties in activity inference. To that end, we propose a data analysis framework to identify user's activity via coupling the mobile phone data with location-based social networks (LBSN) data. The two datasets are integrated into a Bayesian inference module, considering people's circadian rhythms in both time and space. Specifically, the framework considers the pattern of arrival time to each type of facility and the spatial distribution of facilities. The former can be observed from the LBSN Data and the latter is provided by the points of interest (POIs) dataset. Taking Shanghai as an example, we reconstruct the activity chains of 1,000,000 active mobile phone users and analyze the temporal and spatial characteristics of each activity type. We assess the results with some official surveys and a real-world check-in dataset collected in Shanghai, indicating that the proposed method can capture and analyze human activities effectively. Next, we cluster users' inferred activity chains with a topic model to understand the behavior of different groups of users. This data analysis framework provides an example of reconstructing and understanding the activity of the population at an urban scale with big data fusion.

cs.SI↗
arXiv2023

WeLM: A Well-Read Pre-trained Language Model for Chinese

Large Language Models pre-trained with self-supervised learning have demonstrated impressive zero-shot generalization capabilities on a wide spectrum of tasks. In this work, we present WeLM: a well-read pre-trained language model for Chinese that is able to seamlessly perform different types of tasks with zero or few-shot demonstrations. WeLM is trained with 10B parameters by "reading" a curated high-quality corpus covering a wide range of topics. We show that WeLM is equipped with broad knowledge on various domains and languages. On 18 monolingual (Chinese) tasks, WeLM can significantly outperform existing pre-trained models with similar sizes and match the performance of models up to 25 times larger. WeLM also exhibits strong capabilities in multi-lingual and code-switching understanding, outperforming existing multilingual language models pre-trained on 30 languages. Furthermore, We collected human-written prompts for a large set of supervised datasets in Chinese and fine-tuned WeLM with multi-prompted training. The resulting model can attain strong generalization on unseen types of tasks and outperform the unsupervised WeLM in zero-shot learning. Finally, we demonstrate that WeLM has basic skills at explaining and calibrating the decisions from itself, which can be promising directions for future research. Our models can be applied from https://welm.weixin.qq.com/docs/api/.

cs.CL↗
arXiv2023

Adaptive Sparse Pairwise Loss for Object Re-Identification

Object re-identification (ReID) aims to find instances with the same identity as the given probe from a large gallery. Pairwise losses play an important role in training a strong ReID network. Existing pairwise losses densely exploit each instance as an anchor and sample its triplets in a mini-batch. This dense sampling mechanism inevitably introduces positive pairs that share few visual similarities, which can be harmful to the training. To address this problem, we propose a novel loss paradigm termed Sparse Pairwise (SP) loss that only leverages few appropriate pairs for each class in a mini-batch, and empirically demonstrate that it is sufficient for the ReID tasks. Based on the proposed loss framework, we propose an adaptive positive mining strategy that can dynamically adapt to diverse intra-class variations. Extensive experiments show that SP loss and its adaptive variant AdaSP loss outperform other pairwise losses, and achieve state-of-the-art performance across several ReID benchmarks. Code is available at https://github.com/Astaxanthin/AdaSP.

cs.CV↗