Sen Wang

Shihao Zou, Yuxuan Mu, Wei Ji et al.

Event camera, as an asynchronous vision sensor capturing scene dynamics, presents new opportunities for highly efficient 3D human pose tracking. Existing approaches typically adopt modern-day Artificial Neural Networks (ANNs), such as CNNs or Transformer, where sparse events are converted into dense images or paired with additional gray-scale images as input. Such practices, however, ignore the inherent sparsity of events, resulting in redundant computations, increased energy consumption, and potentially degraded performance. Motivated by these observations, we introduce the first sparse Spiking Neural Networks (SNNs) framework for 3D human pose tracking based solely on events. Our approach eliminates the need to convert sparse data to dense formats or incorporate additional images, thereby fully exploiting the innate sparsity of input events. Central to our framework is a novel Spiking Spatiotemporal Transformer, which enables bi-directional spatiotemporal fusion of spike pose features and provides a guaranteed similarity measurement between binary spike features in spiking attention. Moreover, we have constructed a large-scale synthetic dataset, SynEventHPD, that features a broad and diverse set of 3D human motions, as well as much longer hours of event streams. Empirical experiments demonstrate the superiority of our approach over existing state-of-the-art (SOTA) ANN-based methods, requiring only 19.1% FLOPs and 3.6% energy cost. Furthermore, our approach outperforms existing SNN-based benchmarks in this task, highlighting the effectiveness of our proposed SNN framework. The dataset will be released upon acceptance, and code can be found at https://github.com/JimmyZou/HumanPoseTracking_SNN.

Yudong Chen, Sen Wang, Jiajun Liu et al.

In knowledge distillation, previous feature distillation methods mainly focus on the design of loss functions and the selection of the distilled layers, while the effect of the feature projector between the student and the teacher remains under-explored. In this paper, we first discuss a plausible mechanism of the projector with empirical evidence and then propose a new feature distillation method based on a projector ensemble for further performance improvement. We observe that the student network benefits from a projector even if the feature dimensions of the student and the teacher are the same. Training a student backbone without a projector can be considered as a multi-task learning process, namely achieving discriminative feature extraction for classification and feature matching between the student and the teacher for distillation at the same time. We hypothesize and empirically verify that without a projector, the student network tends to overfit the teacher's feature distributions despite having different architecture and weights initialization. This leads to degradation on the quality of the student's deep features that are eventually used in classification. Adding a projector, on the other hand, disentangles the two learning tasks and helps the student network to focus better on the main feature extraction task while still being able to utilize teacher features as a guidance through the projector. Motivated by the positive effect of the projector in feature distillation, we propose an ensemble of projectors to further improve the quality of student features. Experimental results on different datasets with a series of teacher-student pairs illustrate the effectiveness of the proposed method.

Xuwei Xu, Changlin Li, Yudong Chen et al.

Vision Transformers (ViTs) have demonstrated outstanding performance in computer vision tasks, yet their high computational complexity prevents their deployment in computing resource-constrained environments. Various token pruning techniques have been introduced to alleviate the high computational burden of ViTs by dynamically dropping image tokens. However, some undesirable pruning at early stages may result in permanent loss of image information in subsequent layers, consequently hindering model performance. To address this problem, we propose IdleViT, a dynamic token-idle-based method that achieves an excellent trade-off between performance and efficiency. Specifically, in each layer, IdleViT selects a subset of the image tokens to participate in computations while keeping the rest of the tokens idle and directly passing them to this layer's output. By allowing the idle tokens to be re-selected in the following layers, IdleViT mitigates the negative impact of improper pruning in the early stages. Furthermore, inspired by the normalized graph cut, we devise a token cut loss on the attention map as regularization to improve IdleViT's token selection ability. Our method is simple yet effective and can be extended to pyramid ViTs since no token is completely dropped. Extensive experimental results on various ViT architectures have shown that IdleViT can diminish the complexity of pretrained ViTs by up to 33\% with no more than 0.2\% accuracy decrease on ImageNet, after finetuning for only 30 epochs. Notably, when the keep ratio is 0.5, IdleViT outperforms the state-of-the-art EViT on DeiT-S by 0.5\% higher accuracy and even faster inference speed. The source code is available in the supplementary material.

Yi Zhang, Ruihong Qiu, Jiajun Liu et al.

Offline reinforcement learning (RL) is an effective tool for real-world recommender systems with its capacity to model the dynamic interest of users and its interactive nature. Most existing offline RL recommender systems focus on model-based RL through learning a world model from offline data and building the recommendation policy by interacting with this model. Although these methods have made progress in the recommendation performance, the effectiveness of model-based offline RL methods is often constrained by the accuracy of the estimation of the reward model and the model uncertainties, primarily due to the extreme discrepancy between offline logged data and real-world data in user interactions with online platforms. To fill this gap, a more accurate reward model and uncertainty estimation are needed for the model-based RL methods. In this paper, a novel model-based Reward Shaping in Offline Reinforcement Learning for Recommender Systems, ROLeR, is proposed for reward and uncertainty estimation in recommendation systems. Specifically, a non-parametric reward shaping method is designed to refine the reward model. In addition, a flexible and more representative uncertainty penalty is designed to fit the needs of recommendation systems. Extensive experiments conducted on four benchmark datasets showcase that ROLeR achieves state-of-the-art performance compared with existing baselines. The source code can be downloaded at https://github.com/ArronDZhang/ROLeR.

Zixin Wang, Yadan Luo, Zhi Chen et al.

The prevalence of domain adaptive semantic segmentation has prompted concerns regarding source domain data leakage, where private information from the source domain could inadvertently be exposed in the target domain. To circumvent the requirement for source data, source-free domain adaptation has emerged as a viable solution that leverages self-training methods to pseudo-label high-confidence regions and adapt the model to the target data. However, the confidence scores obtained are often highly biased due to over-confidence and class-imbalance issues, which render both model selection and optimization problematic. In this paper, we propose a novel calibration-guided source-free domain adaptive semantic segmentation (Cal-SFDA) framework. The core idea is to estimate the expected calibration error (ECE) from the segmentation predictions, serving as a strong indicator of the model's generalization capability to the unlabeled target domain. The estimated ECE scores, in turn, assist the model training and fair selection in both source training and target adaptation stages. During model pre-training on the source domain, we ensure the differentiability of the ECE objective by leveraging the LogSumExp trick and using ECE scores to select the best source checkpoints for adaptation. To enable ECE estimation on the target domain without requiring labels, we train a value net for ECE estimation and apply statistic warm-up on its BatchNorm layers for stability. The estimated ECE scores assist in determining the reliability of prediction and enable class-balanced pseudo-labeling by positively guiding the adaptation progress and inhibiting potential error accumulation. Extensive experiments on two widely-used synthetic-to-real transfer tasks show that the proposed approach surpasses previous state-of-the-art by up to 5.25% of mIoU with fair model selection criteria.

Bingqing Zhang, Sen Wang, Yifan Liu et al.

Current video object detection (VOD) models often encounter issues with over-aggregation due to redundant aggregation strategies, which perform feature aggregation on every frame. This results in suboptimal performance and increased computational complexity. In this work, we propose an image-level Object Detection Difficulty (ODD) metric to quantify the difficulty of detecting objects in a given image. The derived ODD scores can be used in the VOD process to mitigate over-aggregation. Specifically, we train an ODD predictor as an auxiliary head of a still-image object detector to compute the ODD score for each image based on the discrepancies between detection results and ground-truth bounding boxes. The ODD score enhances the VOD system in two ways: 1) it enables the VOD system to select superior global reference frames, thereby improving overall accuracy; and 2) it serves as an indicator in the newly designed ODD Scheduler to eliminate the aggregation of frames that are easy to detect, thus accelerating the VOD process. Comprehensive experiments demonstrate that, when utilized for selecting global reference frames, ODD-VOD consistently enhances the accuracy of Global-frame-based VOD models. When employed for acceleration, ODD-VOD consistently improves the frames per second (FPS) by an average of 73.3% across 8 different VOD models without sacrificing accuracy. When combined, ODD-VOD attains state-of-the-art performance when competing with many VOD methods in both accuracy and speed. Our work represents a significant advancement towards making VOD more practical for real-world applications.

James Pak, Jyun-Yu Jiang, Fan Zhang et al.

Scaling a Search Conversion Rate (CVR) prediction model, especially in high-traffic environments, presents a challenge: superior model quality needs to be balanced with strict constraints on training cost and serving latency. This paper details an effective approach for scaling modern search CVR prediction models. We begin with an empirical study to understand the scaling performance of search CVR models, analyzing how quality improves as we scale three key factors of model backbone computation, the size of embedding parameters, and the volume of training data. We use a large-scale production dataset, comprising over a year of customer interaction logs from a high-traffic e-commerce platform, to evaluate the scalability of several state-of-the-art architectures and their ensembles. Our key findings are: (1) selecting the right backbone and scaling factors is crucial; (2) the impact of scaling backbone, embedding, and data is largely independent and additive, which has implications for more efficient scaling exploration; (3) a streamlined warmstart strategy can accelerate training iterations while simplifying new updates; (4) inference optimization strategies such as decoupled graph execution and dynamic batching can enable low-latency GPU serving even for high-capacity models. Compared to a baseline of a pre-scaling production model, we ultimately deployed a model trained on 2.5x larger training data with 8x more inference compute while having minimal latency impact. Online A/B tests also demonstrate that our launches achieved a combined +2.6% gain in a key metric of search conversion rate.

Haodong Hong, Sen Wang, Zi Huang et al.

Real-world navigation often involves dealing with unexpected obstructions such as closed doors, moved objects, and unpredictable entities. However, mainstream Vision-and-Language Navigation (VLN) tasks typically assume instructions perfectly align with the fixed and predefined navigation graphs without any obstructions. This assumption overlooks potential discrepancies in actual navigation graphs and given instructions, which can cause major failures for both indoor and outdoor agents. To address this issue, we integrate diverse obstructions into the R2R dataset by modifying both the navigation graphs and visual observations, introducing an innovative dataset and task, R2R with UNexpected Obstructions (R2R-UNO). R2R-UNO contains various types and numbers of path obstructions to generate instruction-reality mismatches for VLN research. Experiments on R2R-UNO reveal that state-of-the-art VLN methods inevitably encounter significant challenges when facing such mismatches, indicating that they rigidly follow instructions rather than navigate adaptively. Therefore, we propose a novel method called ObVLN (Obstructed VLN), which includes a curriculum training strategy and virtual graph construction to help agents effectively adapt to obstructed environments. Empirical results show that ObVLN not only maintains robust performance in unobstructed scenarios but also achieves a substantial performance advantage with unexpected obstructions.

Xueping Peng, Guodong Long, Tao Shen et al.

Sequential diagnosis prediction on the Electronic Health Record (EHR) has been proven crucial for predictive analytics in the medical domain. EHR data, sequential records of a patient's interactions with healthcare systems, has numerous inherent characteristics of temporality, irregularity and data insufficiency. Some recent works train healthcare predictive models by making use of sequential information in EHR data, but they are vulnerable to irregular, temporal EHR data with the states of admission/discharge from hospital, and insufficient data. To mitigate this, we propose an end-to-end robust transformer-based model called SETOR, which exploits neural ordinary differential equation to handle both irregular intervals between a patient's visits with admitted timestamps and length of stay in each visit, to alleviate the limitation of insufficient data by integrating medical ontology, and to capture the dependencies between the patient's visits by employing multi-layer transformer blocks. Experiments conducted on two real-world healthcare datasets show that, our sequential diagnoses prediction model SETOR not only achieves better predictive results than previous state-of-the-art approaches, irrespective of sufficient or insufficient training data, but also derives more interpretable embeddings of medical codes. The experimental codes are available at the GitHub repository (https://github.com/Xueping/SETOR).

Xueping Peng, Guodong Long, Tao Shen et al.

Understanding patients' journeys in healthcare system is a fundamental prepositive task for a broad range of AI-based healthcare applications. This task aims to learn an informative representation that can comprehensively encode hidden dependencies among medical events and its inner entities, and then the use of encoding outputs can greatly benefit the downstream application-driven tasks. A patient journey is a sequence of electronic health records (EHRs) over time that is organized at multiple levels: patient, visits and medical codes. The key challenge of patient journey understanding is to design an effective encoding mechanism which can properly tackle the aforementioned multi-level structured patient journey data with temporal sequential visits and a set of medical codes. This paper proposes a novel self-attention mechanism that can simultaneously capture the contextual and temporal relationships hidden in patient journeys. A multi-level self-attention network (MusaNet) is specifically designed to learn the representations of patient journeys that is used to be a long sequence of activities. The MusaNet is trained in end-to-end manner using the training data derived from EHRs. We evaluated the efficacy of our method on two medical application tasks with real-world benchmark datasets. The results have demonstrated the proposed MusaNet produces higher-quality representations than state-of-the-art baseline methods. The source code is available in https://github.com/xueping/MusaNet.

Team Seawead, Ceyuan Yang, Zhijie Lin et al.

This technical report presents a cost-efficient strategy for training a video generation foundation model. We present a mid-sized research model with approximately 7 billion parameters (7B) called Seaweed-7B trained from scratch using 665,000 H100 GPU hours. Despite being trained with moderate computational resources, Seaweed-7B demonstrates highly competitive performance compared to contemporary video generation models of much larger size. Design choices are especially crucial in a resource-constrained setting. This technical report highlights the key design decisions that enhance the performance of the medium-sized diffusion model. Empirically, we make two observations: (1) Seaweed-7B achieves performance comparable to, or even surpasses, larger models trained on substantially greater GPU resources, and (2) our model, which exhibits strong generalization ability, can be effectively adapted across a wide range of downstream applications either by lightweight fine-tuning or continue training. See the project page at https://seaweed.video/

Zixin Wang, Dong Gong, Sen Wang et al.

Contrastive Language-Image Pretraining (CLIP) excels at learning generalizable image representations but often falls short in zero-shot inference on certain downstream datasets. Test-time adaptation (TTA) mitigates this issue by adjusting components like normalization layers or context prompts, yet it typically requires large batch sizes and extensive augmentations, leading to high computational costs. This raises a key question: Can VLMs' performance drop in specific test cases be mitigated through efficient, training-free approaches? To explore the solution, we investigate token condensation (TC) techniques, originally designed to enhance vision transformer efficiency by refining token usage during inference. We observe that informative tokens improve visual-text alignment in VLMs like CLIP on unseen datasets. However, existing TC methods often fail to maintain in-distribution performance when reducing tokens, prompting us to ask: How can we transform TC into an effective ``free-lunch'' adaptation strategy for VLMs? To address this, we propose Token Condensation as Adaptation (TCA), a training-free adaptation method that takes a step beyond standard TC. Rather than passively discarding tokens, TCA condenses token representation by introducing reservoir-based domain anchor tokens for information-preserving token reduction and logits correction. TCA achieves up to a 21.4% performance improvement over the strongest baseline on cross-dataset benchmark and the CIFAR-100-Corrupted dataset while reducing GFLOPs by 12.2% to 48.9%, with minimal hyperparameter dependency on both CLIP and SigLIP series.

Haodong Hong, Sen Wang, Zi Huang et al.

Current Vision-and-Language Navigation (VLN) tasks mainly employ textual instructions to guide agents. However, being inherently abstract, the same textual instruction can be associated with different visual signals, causing severe ambiguity and limiting the transfer of prior knowledge in the vision domain from the user to the agent. To fill this gap, we propose Vision-and-Language Navigation with Multi-modal Prompts (VLN-MP), a novel task augmenting traditional VLN by integrating both natural language and images in instructions. VLN-MP not only maintains backward compatibility by effectively handling text-only prompts but also consistently shows advantages with different quantities and relevance of visual prompts. Possible forms of visual prompts include both exact and similar object images, providing adaptability and versatility in diverse navigation scenarios. To evaluate VLN-MP under a unified framework, we implement a new benchmark that offers: (1) a training-free pipeline to transform textual instructions into multi-modal forms with landmark images; (2) diverse datasets with multi-modal instructions for different downstream tasks; (3) a novel module designed to process various image prompts for seamless integration with state-of-the-art VLN models. Extensive experiments on four VLN benchmarks (R2R, RxR, REVERIE, CVDN) show that incorporating visual prompts significantly boosts navigation performance. While maintaining efficiency with text-only prompts, VLN-MP enables agents to navigate in the pre-explore setting and outperform text-based models, showing its broader applicability.

Chuan Guo, Xinxin Zuo, Sen Wang et al.

We aim to tackle the interesting yet challenging problem of generating videos of diverse and natural human motions from prescribed action categories. The key issue lies in the ability to synthesize multiple distinct motion sequences that are realistic in their visual appearances. It is achieved in this paper by a two-step process that maintains internal 3D pose and shape representations, action2motion and motion2video. Action2motion stochastically generates plausible 3D pose sequences of a prescribed action category, which are processed and rendered by motion2video to form 2D videos. Specifically, the Lie algebraic theory is engaged in representing natural human motions following the physical law of human kinematics; a temporal variational auto-encoder (VAE) is developed that encourages diversity of output motions. Moreover, given an additional input image of a clothed human character, an entire pipeline is proposed to extract his/her 3D detailed shape, and to render in videos the plausible motions from different views. This is realized by improving existing methods to extract 3D human shapes and textures from single 2D images, rigging, animating, and rendering to form 2D videos of human motions. It also necessitates the curation and reannotation of 3D human motion datasets for training purpose. Thorough empirical experiments including ablation study, qualitative and quantitative evaluations manifest the applicability of our approach, and demonstrate its competitiveness in addressing related tasks, where components of our approach are compared favorably to the state-of-the-arts.

Xueping Peng, Guodong Long, Tao Shen et al.

Representation learning on electronic health records (EHRs) plays a vital role in downstream medical prediction tasks. Although natural language processing techniques, such as recurrent neural networks, and self-attention, have been adapted for learning medical representations from hierarchical, time-stamped EHR data, they often struggle when either general or task-specific data are limited. Recent efforts have attempted to mitigate this challenge by incorporating medical ontologies (i.e., knowledge graphs) into self-supervised tasks like diagnosis prediction. However, two main issues remain: (1) small and uniform ontologies that lack diversity for robust learning, and (2) insufficient attention to the critical contexts or dependencies underlying patient journeys, which could further enhance ontology-based learning. To address these gaps, we propose MIPO (Mutual Integration of Patient Journey and Medical Ontology), a robust end-to-end framework that employs a Transformer-based architecture for representation learning. MIPO emphasizes task-specific representation learning through a sequential diagnosis prediction task, while also incorporating an ontology-based disease-typing task. A graph-embedding module is introduced to integrate information from patient visit records, thus alleviating data insufficiency. This setup creates a mutually reinforcing loop, where both patient-journey embedding and ontology embedding benefit from each other. We validate MIPO on two real-world benchmark datasets, showing that it consistently outperforms baseline methods under both sufficient and limited data conditions. Furthermore, the resulting diagnosis embeddings offer improved interpretability, underscoring the promise of MIPO for real-world healthcare applications.

Xueping Peng, Guodong Long, Tao Shen et al.

Electronic health records (EHRs) are longitudinal records of a patient's interactions with healthcare systems. A patient's EHR data is organized as a three-level hierarchy from top to bottom: patient journey - all the experiences of diagnoses and treatments over a period of time; individual visit - a set of medical codes in a particular visit; and medical code - a specific record in the form of medical codes. As EHRs begin to amass in millions, the potential benefits, which these data might hold for medical research and medical outcome prediction, are staggering - including, for example, predicting future admissions to hospitals, diagnosing illnesses or determining the efficacy of medical treatments. Each of these analytics tasks requires a domain knowledge extraction method to transform the hierarchical patient journey into a vector representation for further prediction procedure. The representations should embed a sequence of visits and a set of medical codes with a specific timestamp, which are crucial to any downstream prediction tasks. Hence, expressively powerful representations are appealing to boost learning performance. To this end, we propose a novel self-attention mechanism that captures the contextual dependency and temporal relationships within a patient's healthcare journey. An end-to-end bidirectional temporal encoder network (BiteNet) then learns representations of the patient's journeys, based solely on the proposed attention mechanism. We have evaluated the effectiveness of our methods on two supervised prediction and two unsupervised clustering tasks with a real-world EHR dataset. The empirical results demonstrate the proposed BiteNet model produces higher-quality representations than state-of-the-art baseline methods.

Lingling Zhang, Xiaojun Chang, Jun Liu et al.

An integral part of video analysis and surveillance is temporal activity detection, which means to simultaneously recognize and localize activities in long untrimmed videos. Currently, the most effective methods of temporal activity detection are based on deep learning, and they typically perform very well with large scale annotated videos for training. However, these methods are limited in real applications due to the unavailable videos about certain activity classes and the time-consuming data annotation. To solve this challenging problem, we propose a novel task setting called zero-shot temporal activity detection (ZSTAD), where activities that have never been seen in training can still be detected. We design an end-to-end deep network based on R-C3D as the architecture for this solution. The proposed network is optimized with an innovative loss function that considers the embeddings of activity labels and their super-classes while learning the common semantics of seen and unseen activities. Experiments on both the THUMOS14 and the Charades datasets show promising performance in terms of detecting unseen activities.

Xueping Peng, Guodong Long, Tao Shen et al.

In longitudinal electronic health records (EHRs), the event records of a patient are distributed over a long period of time and the temporal relations between the events reflect sufficient domain knowledge to benefit prediction tasks such as the rate of inpatient mortality. Medical concept embedding as a feature extraction method that transforms a set of medical concepts with a specific time stamp into a vector, which will be fed into a supervised learning algorithm. The quality of the embedding significantly determines the learning performance over the medical data. In this paper, we propose a medical concept embedding method based on applying a self-attention mechanism to represent each medical concept. We propose a novel attention mechanism which captures the contextual information and temporal relationships between medical concepts. A light-weight neural net, "Temporal Self-Attention Network (TeSAN)", is then proposed to learn medical concept embedding based solely on the proposed attention mechanism. To test the effectiveness of our proposed methods, we have conducted clustering and prediction tasks on two public EHRs datasets comparing TeSAN against five state-of-the-art embedding methods. The experimental results demonstrate that the proposed TeSAN model is superior to all the compared methods. To the best of our knowledge, this work is the first to exploit temporal self-attentive relations between medical events.

Shining Liang, Wanli Zuo, Zhenkun Shi et al.

Mining causality from text is a complex and crucial natural language understanding task corresponding to the human cognition. Existing studies at its solution can be grouped into two primary categories: feature engineering based and neural model based methods. In this paper, we find that the former has incomplete coverage and inherent errors but provide prior knowledge; while the latter leverages context information but causal inference of which is insufficiency. To handle the limitations, we propose a novel causality detection model named MCDN to explicitly model causal reasoning process, and furthermore, to exploit the advantages of both methods. Specifically, we adopt multi-head self-attention to acquire semantic feature at word level and develop the SCRN to infer causality at segment level. To the best of our knowledge, with regards to the causality tasks, this is the first time that the Relation Network is applied. The experimental results show that: 1) the proposed approach performs prominent performance on causality detection; 2) further analysis manifests the effectiveness and robustness of MCDN.

Shaoxiong Ji, Guodong Long, Shirui Pan et al.

The Internet and the Web are being increasingly used in proactive social care to provide people, especially the vulnerable, with a better life and services, and their derived social services generate enormous data. However, the strict protection of privacy makes user's data become an isolated island and limits the predictive performance of standalone clients. To enable effective proactive social care and knowledge sharing within intelligent agents, this paper develops a knowledge transferring framework via model aggregation. Under this framework, distributed clients perform on-device training, and a third-party server integrates multiple clients' models and redistributes to clients for knowledge transferring among users. To improve the generalizability of the knowledge sharing, we further propose a novel model aggregation algorithm, namely the average difference descent aggregation (AvgDiffAgg for short). In particular, to evaluate the effectiveness of the learning algorithm, we use a case study on the early detection and prevention of suicidal ideation, and the experiment results on four datasets derived from social communities demonstrate the effectiveness of the proposed learning method.

Shuai Zhang, Lina Yao, Aixin Sun et al.

Modeling user-item interaction patterns is an important task for personalized recommendations. Many recommender systems are based on the assumption that there exists a linear relationship between users and items while neglecting the intricacy and non-linearity of real-life historical interactions. In this paper, we propose a neural network based recommendation model (NeuRec) that untangles the complexity of user-item interactions and establishes an integrated network to combine non-linear transformation with latent factors. We further design two variants of NeuRec: user-based NeuRec and item-based NeuRec, by concentrating on different aspects of the interaction matrix. Extensive experiments on four real-world datasets demonstrated their superior performances on personalized ranking task.

Xiang Zhang, Lina Yao, Chaoran Huang et al.

Multimodal wearable sensor data classification plays an important role in ubiquitous computing and has a wide range of applications in scenarios from healthcare to entertainment. However, most existing work in this field employs domain-specific approaches and is thus ineffective in complex sit- uations where multi-modality sensor data are col- lected. Moreover, the wearable sensor data are less informative than the conventional data such as texts or images. In this paper, to improve the adapt- ability of such classification methods across differ- ent application domains, we turn this classification task into a game and apply a deep reinforcement learning scheme to deal with complex situations dynamically. Additionally, we introduce a selective attention mechanism into the reinforcement learn- ing scheme to focus on the crucial dimensions of the data. This mechanism helps to capture extra information from the signal and thus it is able to significantly improve the discriminative power of the classifier. We carry out several experiments on three wearable sensor datasets and demonstrate the competitive performance of the proposed approach compared to several state-of-the-art baselines.

Tao Shen, Tianyi Zhou, Guodong Long et al.

Many natural language processing tasks solely rely on sparse dependencies between a few tokens in a sentence. Soft attention mechanisms show promising performance in modeling local/global dependencies by soft probabilities between every two tokens, but they are not effective and efficient when applied to long sentences. By contrast, hard attention mechanisms directly select a subset of tokens but are difficult and inefficient to train due to their combinatorial nature. In this paper, we integrate both soft and hard attention into one context fusion model, "reinforced self-attention (ReSA)", for the mutual benefit of each other. In ReSA, a hard attention trims a sequence for a soft self-attention to process, while the soft attention feeds reward signals back to facilitate the training of the hard one. For this purpose, we develop a novel hard attention called "reinforced sequence sampling (RSS)", selecting tokens in parallel and trained via policy gradient. Using two RSS modules, ReSA efficiently extracts the sparse dependencies between each pair of selected tokens. We finally propose an RNN/CNN-free sentence-encoding model, "reinforced self-attention network (ReSAN)", solely based on ReSA. It achieves state-of-the-art performance on both Stanford Natural Language Inference (SNLI) and Sentences Involving Compositional Knowledge (SICK) datasets.

Fang Gong, Meng Wang, Haofen Wang et al.

Most of the existing medicine recommendation systems that are mainly based on electronic medical records (EMRs) are significantly assisting doctors to make better clinical decisions benefiting both patients and caregivers. Even though the growth of EMRs is at a lighting fast speed in the era of big data, content limitations in EMRs restrain the existed recommendation systems to reflect relevant medical facts, such as drug-drug interactions. Many medical knowledge graphs that contain drug-related information, such as DrugBank, may give hope for the recommendation systems. However, the direct use of these knowledge graphs in the systems suffers from robustness caused by the incompleteness of the graphs. To address these challenges, we stand on recent advances in graph embedding learning techniques and propose a novel framework, called Safe Medicine Recommendation (SMR), in this paper. Specifically, SMR first constructs a high-quality heterogeneous graph by bridging EMRs (MIMIC-III) and medical knowledge graphs (ICD-9 ontology and DrugBank). Then, SMR jointly embeds diseases, medicines, patients, and their corresponding relations into a shared lower dimensional space. Finally, SMR uses the embeddings to decompose the medicine recommendation into a link prediction process while considering the patient's diagnoses and adverse drug reactions. To our best knowledge, SMR is the first to learn embeddings of a patient-disease-medicine graph for medicine recommendation in the world. Extensive experiments on real datasets are conducted to evaluate the effectiveness of proposed framework.

Dalin Zhang, Lina Yao, Xiang Zhang et al.

Brain-Computer Interface (BCI) is a system empowering humans to communicate with or control the outside world with exclusively brain intentions. Electroencephalography (EEG) based BCIs are promising solutions due to their convenient and portable instruments. Motor imagery EEG (MI-EEG) is a kind of most widely focused EEG signals, which reveals a subjects movement intentions without actual actions. Despite the extensive research of MI-EEG in recent years, it is still challenging to interpret EEG signals effectively due to the massive noises in EEG signals (e.g., low signal noise ratio and incomplete EEG signals), and difficulties in capturing the inconspicuous relationships between EEG signals and certain brain activities. Most existing works either only consider EEG as chain-like sequences neglecting complex dependencies between adjacent signals or performing simple temporal averaging over EEG sequences. In this paper, we introduce both cascade and parallel convolutional recurrent neural network models for precisely identifying human intended movements by effectively learning compositional spatio-temporal representations of raw EEG streams. The proposed models grasp the spatial correlations between physically neighboring EEG signals by converting the chain like EEG sequences into a 2D mesh like hierarchy. An LSTM based recurrent network is able to extract the subtle temporal dependencies of EEG data streams. Extensive experiments on a large-scale MI-EEG dataset (108 subjects, 3,145,160 EEG records) have demonstrated that both models achieve high accuracy near 98.3% and outperform a set of baseline methods and most recent deep learning based EEG recognition models, yielding a significant accuracy increase of 18% in the cross-subject validation scenario.

Meng Wang, Jiaheng Zhang, Jun Liu et al.

Electronic medical records contain multi-format electronic medical data that consist of an abundance of medical knowledge. Facing with patient's symptoms, experienced caregivers make right medical decisions based on their professional knowledge that accurately grasps relationships between symptoms, diagnosis and corresponding treatments. In this paper, we aim to capture these relationships by constructing a large and high-quality heterogenous graph linking patients, diseases, and drugs (PDD) in EMRs. Specifically, we propose a novel framework to extract important medical entities from MIMIC-III (Medical Information Mart for Intensive Care III) and automatically link them with the existing biomedical knowledge graphs, including ICD-9 ontology and DrugBank. The PDD graph presented in this paper is accessible on the Web via the SPARQL endpoint, and provides a pathway for medical discovery and applications, such as effective treatment recommendations.

Shuai Zhang, Lina Yao, Xiwei Xu et al.

In this paper, we present a novel structure, Semi-AutoEncoder, based on AutoEncoder. We generalize it into a hybrid collaborative filtering model for rating prediction as well as personalized top-n recommendations. Experimental results on two real-world datasets demonstrate its state-of-the-art performances.

Ronald Clark, Sen Wang, Hongkai Wen et al.

Localization is a key requirement for mobile robot autonomy and human-robot interaction. Vision-based localization is accurate and flexible, however, it incurs a high computational burden which limits its application on many resource-constrained platforms. In this paper, we address the problem of performing real-time localization in large-scale 3D point cloud maps of ever-growing size. While most systems using multi-modal information reduce localization time by employing side-channel information in a coarse manner (eg. WiFi for a rough prior position estimate), we propose to inter-weave the map with rich sensory data. This multi-modal approach achieves two key goals simultaneously. First, it enables us to harness additional sensory data to localise against a map covering a vast area in real-time; and secondly, it also allows us to roughly localise devices which are not equipped with a camera. The key to our approach is a localization policy based on a sequential Monte Carlo estimator. The localiser uses this policy to attempt point-matching only in nodes where it is likely to succeed, significantly increasing the efficiency of the localization process. The proposed multi-modal localization system is evaluated extensively in a large museum building. The results show that our multi-modal approach not only increases the localization accuracy but significantly reduces computational time.

Sen Wang, Feiping Nie, Xiaojun Chang et al.

Manifold structure learning is often used to exploit geometric information among data in semi-supervised feature learning algorithms. In this paper, we find that local discriminative information is also of importance for semi-supervised feature learning. We propose a method that utilizes both the manifold structure of data and local discriminant information. Specifically, we define a local clique for each data point. The k-Nearest Neighbors (kNN) is used to determine the structural information within each clique. We then employ a variant of Fisher criterion model to each clique for local discriminant evaluation and sum all cliques as global integration into the framework. In this way, local discriminant information is embedded. Labels are also utilized to minimize distances between data from the same class. In addition, we use the kernel method to extend our proposed model and facilitate feature learning in a high-dimensional space after feature mapping. Experimental results show that our method is superior to all other compared methods over a number of datasets.

Sen Wang, Feiping Nie, Xiaojun Chang et al.

Unsupervised feature selection has been always attracting research attention in the communities of machine learning and data mining for decades. In this paper, we propose an unsupervised feature selection method seeking a feature coefficient matrix to select the most distinctive features. Specifically, our proposed algorithm integrates the Maximum Margin Criterion with a sparsity-based model into a joint framework, where the class margin and feature correlation are taken into account at the same time. To maximize the total data separability while preserving minimized within-class scatter simultaneously, we propose to embed Kmeans into the framework generating pseudo class label information in a scenario of unsupervised feature selection. Meanwhile, a sparsity-based model, ` 2 ,p-norm, is imposed to the regularization term to effectively discover the sparse structures of the feature coefficient matrix. In this way, noisy and irrelevant features are removed by ruling out those features whose corresponding coefficients are zeros. To alleviate the local optimum problem that is caused by random initializations of K-means, a convergence guaranteed algorithm with an updating strategy for the clustering indicator matrix, is proposed to iteractively chase the optimal solution. Performance evaluation is extensively conducted over six benchmark data sets. From plenty of experimental results, it is demonstrated that our method has superior performance against all other compared approaches.

Xiaojun Chang, Feiping Nie, Sen Wang et al.

In many real-world applications, data are represented by matrices or high-order tensors. Despite the promising performance, the existing two-dimensional discriminant analysis algorithms employ a single projection model to exploit the discriminant information for projection, making the model less flexible. In this paper, we propose a novel Compound Rank-k Projection (CRP) algorithm for bilinear analysis. CRP deals with matrices directly without transforming them into vectors, and it therefore preserves the correlations within the matrix and decreases the computation complexity. Different from the existing two dimensional discriminant analysis algorithms, objective function values of CRP increase monotonically.In addition, CRP utilizes multiple rank-k projection models to enable a larger search space in which the optimal solution can be found. In this way, the discriminant ability is enhanced.