HimGNN: a novel hierarchical molecular graph representation learning framework for property prediction.

Accurate prediction of molecular properties is an important topic in drug discovery. Recent works have developed various representation schemes for molecular structures to capture different chemical information in molecules. The atom and motif can be viewed as hierarchical molecular structures that are widely used for learning molecular representations to predict chemical properties. Previous works have attempted to exploit both atom and motif to address the problem of information loss in single representation learning for various tasks. To further fuse such hierarchical information, the correspondence between learned chemical features from different molecular structures should be considered. Herein, we propose a novel framework for molecular property prediction, called hierarchical molecular graph neural networks (HimGNN). HimGNN learns hierarchical topology representations by applying graph neural networks on atom- and motif-based graphs. In order to boost the representational power of the motif feature, we design a Transformer-based local augmentation module to enrich motif features by introducing heterogeneous atom information in motif representation learning. Besides, we focus on the molecular hierarchical relationship and propose a simple yet effective rescaling module, called contextual self-rescaling, that adaptively recalibrates molecular representations by explicitly modelling interdependencies between atom and motif features. Extensive computational experiments demonstrate that HimGNN can achieve promising performances over state-of-the-art baselines on both classification and regression tasks in molecular property prediction.

META-DDIE: predicting drug-drug interaction events with few-shot learning.

Drug-drug interactions (DDIs) are one of the major concerns in pharmaceutical research, and a number of computational methods have been developed to predict whether two drugs interact or not. Recently, more attention has been paid to events caused by the DDIs, which is more useful for investigating the mechanism hidden behind the combined drug usage or adverse reactions. However, some rare events may only have few examples, hindering them from being precisely predicted. To address the above issues, we present a few-shot computational method named META-DDIE, which consists of a representation module and a comparing module, to predict DDI events. We collect drug chemical structures and DDIs from DrugBank, and categorize DDI events into hundreds of types using a standard pipeline. META-DDIE uses the structures of drugs as input and learns the interpretable representations of DDIs through the representation module. Then, the model uses the comparing module to predict whether two representations are similar, and finally predicts DDI events with few labeled examples. In the computational experiments, META-DDIE outperforms several baseline methods and especially enhances the predictive capability for rare events. Moreover, META-DDIE helps to identify the key factors that may cause DDI events and reveal the relationship among different events.

Mesenchymal stem cells protect against TBI-induced pyroptosis in vivo and in vitro through TSG-6.

BACKGROUND: Pyroptosis, especially microglial pyroptosis, may play an important role in central nervous system pathologies, including traumatic brain injury (TBI). Transplantation of mesenchymal stem cells (MSCs), such as human umbilical cord MSCs (hUMSCs), has been a focus of brain injury treatment. Recently, MSCs have been found to play a role in many diseases by regulating the pyroptosis pathway. However, the effect of MSC transplantation on pyroptosis following TBI remains unknown. Tumor necrosis factor α stimulated gene 6/protein (TSG-6), a potent anti-inflammatory factor expressed in many cell types including MSCs, plays an anti-inflammatory role in many diseases; however, the effect of TSG-6 secreted by MSCs on pyroptosis remains unclear. METHODS: Mice were subjected to controlled cortical impact injury in vivo. To assess the time course of pyroptosis after TBI, brains of TBI mice were collected at different time points. To study the effect of TSG-6 secreted by hUMSCs in regulating pyroptosis, normal hUMSCs, sh-TSG-6 hUMSCs, or different concentrations of rmTSG-6 were injected intracerebroventricularly into mice 4 h after TBI. Neurological deficits, double immunofluorescence staining, presence of inflammatory factors, cell apoptosis, and pyroptosis were assessed. In vitro, we investigated the anti-pyroptosis effects of hUMSCs and TSG-6 in a lipopolysaccharide/ATP-induced BV2 microglial pyroptosis model. RESULTS: In TBI mice, the co-localization of Iba-1 (marking microglia/macrophages) with NLRP3/Caspase-1 p20/GSDMD was distinctly observed at 48 h. In vivo, hUMSC transplantation or treatment with rmTSG-6 in TBI mice significantly improved neurological deficits, reduced inflammatory cytokine expression, and inhibited both NLRP3/Caspase-1 p20/GSDMD expression and microglial pyroptosis in the cerebral cortices of TBI mice. However, the therapeutic effect of hUMSCs on TBI mice was reduced by the inhibition of TSG-6 expression in hUMSCs. In vitro, lipopolysaccharide/ATP-induced BV2 microglial pyroptosis was inhibited by co-culture with hUMSCs or with rmTSG-6. However, the inhibitory effect of hUMSCs on BV2 microglial pyroptosis was significantly reduced by TSG-6-shRNA transfection. CONCLUSION: In TBI mice, microglial pyroptosis was observed. Both in vivo and in vitro, hUMSCs inhibited pyroptosis, particularly microglial pyroptosis, by regulating the NLRP3/Caspase-1/GSDMD signaling pathway via TSG-6. Video Abstract.

Tolerant Self-Distillation for image classification.

Deep neural networks tend to suffer from the overfitting issue when the training data are not enough. In this paper, we introduce two metrics from the intra-class distribution of correct-predicted and incorrect-predicted samples to provide a new perspective on the overfitting issue. Based on it, we propose a knowledge distillation approach without pretraining a teacher model in advance named Tolerant Self-Distillation (TSD) for alleviating the overfitting issue. It introduces an online updating memory and selectively stores the class predictions of the samples from the past iterations, making it possible to distill knowledge across the iterations. Specifically, the class predictions stored in the memory bank serve as the soft labels for supervising the samples from the same class for the current iteration in a reverse way, i.e. the correct-predicted samples are supervised with the incorrect predictions while the incorrect-predicted samples are supervised with the correct predictions. Consequently, the premature convergence issue caused by the over-confident samples would be mitigated, which helps the model to converge to a better local optimum. Extensive experimental results on several image classification benchmarks, including small-scale, large-scale, and fine-grained datasets, demonstrate the superiority of the proposed TSD.

Clustering on heterogeneous IoT information network based on meta path.

As the Internet and Internet of Things (IoT) continue to develop, Heterogeneous Information Networks (HIN) have formed complex interaction relationships among data objects. These relationships are represented by various types of edges (meta-paths) that contain rich semantic information. In the context of IoT data applications, the widespread adoption of Trigger-Action Patterns makes the management and analysis of heterogeneous data particularly important. This study proposes a meta-path-based clustering method for heterogeneous IoT data called I-RankClus, which aims to improve the modeling and analysis efficiency of IoT data. By combining ranking with clustering algorithms, the PageRank algorithm was used to calculate the intraclass influence of objects in the network. The HITS algorithm then transfers the influence to the core objects, thereby optimizing the classification of objects during the clustering process. The I-RankClus algorithm does not process each meta-path individually, but instead integrates multiple meta-paths to enhance the interpretability and clustering performance of the model. The experimental results show that the I-RankClus algorithm can process complex IoT datasets more effectively than traditional clustering methods and provide more accurate clustering outcomes. Furthermore, through a detailed analysis of meta-paths, this study explored the influence and importance of different meta-paths, thereby validating the effectiveness of the algorithm. Overall, the research presented in this paper not only improves the application effects of HINs in IoT data analysis but also provides valuable methods and insights for future network data processing.

Small-molecule Molephantin induces apoptosis and mitophagy flux blockage through ROS production in glioblastoma.

Glioblastoma (GBM), one of the most malignant brain tumors in the world, has limited treatment options and a dismal survival rate. Effective and safe disease-modifying drugs for glioblastoma are urgently needed. Here, we identified a small molecule, Molephantin (EM-5), effectively penetrated the blood-brain barrier (BBB) and demonstrated notable antitumor effects against GBM with good safety profiles both in vitro and in vivo. Mechanistically, EM-5 not only inhibits the proliferation and invasion of GBM cells but also induces cell apoptosis through the reactive oxygen species (ROS)-mediated PI3K/Akt/mTOR pathway. Furthermore, EM-5 causes mitochondrial dysfunction and blocks mitophagy flux by impeding the fusion of mitophagosomes with lysosomes. It is noteworthy that EM-5 does not interfere with the initiation of autophagosome formation or lysosomal function. Additionally, the mitophagy flux blockage caused by EM-5 was driven by the accumulation of intracellular ROS. In vivo, EM-5 exhibited significant efficacy in suppressing tumor growth in a xenograft model. Collectively, our findings not only identified EM-5 as a promising, effective, and safe lead compound for treating GBM but also uncovered its underlying mechanisms from the perspective of apoptosis and mitophagy.

Knowledge Distillation Classifier Generation Network for Zero-Shot Learning.

In this article, we present a conceptually simple but effective framework called knowledge distillation classifier generation network (KDCGN) for zero-shot learning (ZSL), where the learning agent requires recognizing unseen classes that have no visual data for training. Different from the existing generative approaches that synthesize visual features for unseen classifiers' learning, the proposed framework directly generates classifiers for unseen classes conditioned on the corresponding class-level semantics. To ensure the generated classifiers to be discriminative to the visual features, we borrow the knowledge distillation idea to both supervise the classifier generation and distill the knowledge with, respectively, the visual classifiers and soft targets trained from a traditional classification network. Under this framework, we develop two, respectively, strategies, i.e., class augmentation and semantics guidance, to facilitate the supervision process from the perspectives of improving visual classifiers. Specifically, the class augmentation strategy incorporates some additional categories to train the visual classifiers, which regularizes the visual classifier weights to be compact, under supervision of which the generated classifiers will be more discriminative. The semantics-guidance strategy encodes the class semantics into the visual classifiers, which would facilitate the supervision process by minimizing the differences between the generated and the real-visual classifiers. To evaluate the effectiveness of the proposed framework, we have conducted extensive experiments on five datasets in image classification, i.e., AwA1, AwA2, CUB, FLO, and APY. Experimental results show that the proposed approach performs best in the traditional ZSL task and achieves a significant performance improvement on four out of the five datasets in the generalized ZSL task.

Balancing Feature Alignment and Uniformity for Few-Shot Classification.

In Few-Shot Learning (FSL), the objective is to correctly recognize new samples from novel classes with only a few available samples per class. Existing methods in FSL primarily focus on learning transferable knowledge from base classes by maximizing the information between feature representations and their corresponding labels. However, this approach may suffer from the "supervision collapse" issue, which arises due to a bias towards the base classes. In this paper, we propose a solution to address this issue by preserving the intrinsic structure of the data and enabling the learning of a generalized model for the novel classes. Following the InfoMax principle, our approach maximizes two types of mutual information (MI): between the samples and their feature representations, and between the feature representations and their class labels. This allows us to strike a balance between discrimination (capturing class-specific information) and generalization (capturing common characteristics across different classes) in the feature representations. To achieve this, we adopt a unified framework that perturbs the feature embedding space using two low-bias estimators. The first estimator maximizes the MI between a pair of intra-class samples, while the second estimator maximizes the MI between a sample and its augmented views. This framework effectively combines knowledge distillation between class-wise pairs and enlarges the diversity in feature representations. By conducting extensive experiments on popular FSL benchmarks, our proposed approach achieves comparable performances with state-of-the-art competitors. For example, we achieved an accuracy of 69.53% on the miniImageNet dataset and 77.06% on the CIFAR-FS dataset for the 5-way 1-shot task.

Enhancing Drug-Drug Interaction Prediction Using Deep Attention Neural Networks.

Drug-drug interactions are one of the main concerns in drug discovery. Accurate prediction of drug-drug interactions plays a key role in increasing the efficiency of drug research and safety when multiple drugs are co-prescribed. With various data sources that describe the relationships and properties between drugs, the comprehensive approach that integrates multiple data sources would be considerably effective in making high-accuracy prediction. In this paper, we propose a Deep Attention Neural Network based Drug-Drug Interaction prediction framework, abbreviated as DANN-DDI, to predict unobserved drug-drug interactions. First, we construct multiple drug feature networks and learn drug representations from these networks using the graph embedding method; then, we concatenate the learned drug embeddings and design an attention neural network to learn representations of drug-drug pairs; finally, we adopt a deep neural network to accurately predict drug-drug interactions. The experimental results demonstrate that our model DANN-DDI has improved prediction performance compared with state-of-the-art methods. Moreover, the proposed model can predict novel drug-drug interactions and drug-drug interaction-associated events.

MCPIP-1 knockdown enhances endothelial colony-forming cell angiogenesis via the TFRC/AKT/mTOR signaling pathway in the ischemic penumbra of MCAO mice.

Cerebral ischemia is a serious disease characterized by brain tissue ischemia and hypoxic necrosis caused by the blockage of blood vessels within the central nervous system. Although stem cell therapy is a promising approach for treating ischemic stroke, the inflammatory, oxidative, and hypoxic environment generated by cerebral ischemia greatly reduces the survival and therapeutic effects of transplanted stem cells. Endothelial colony-forming cells (ECFCs) are a class of precursor cells with strong proliferative potential that can migrate and differentiate directly into mature vascular endothelial cells. Consequently, ECFCs can exert significant therapeutic and reparative effects in diseases associated with vascular injury. Monocyte chemoattractant protein-induced protein 1 (MCPIP-1) exerts multiple biological effects; however, no studies have yet reported its role in the angiogenic function of ECFCs. In this study, we performed Proteome Profiler™ Human Angiogenesis Antibody arrays and tandem mass tag protein profiling to investigate the effect of MCPIP-1 on ECFCs. We demonstrated that MCPIP-1 knockdown enhanced the proliferation, migration, and in vivo and in vitro angiogenic capacity of ECFCs by upregulating the transferrin receptor-activated AKT/m-TOR signaling pathway to promote cellular trophic factor secretion. Furthermore, we found that the lateral ventricular transplantation of ECFCs with lentiviral MCPIP-1 knockdown into mice with middle cerebral artery occlusion increased serum vacular endothelial growth factor(VEGF), angiopoietin-1, and HIF-1a levels, enhanced neovascularization and neurogenesis in the ischemic penumbra, reduced the size of cerebral infarcts, and promoted neurological recovery. Together, these findings suggest new avenues for enhancing the therapeutic efficacy of ECFCs.

Sparsity-control ternary weight networks.

Deep neural networks (DNNs) have been widely and successfully applied to various applications, but they require large amounts of memory and computational power. This severely restricts their deployment on resource-limited devices. To address this issue, many efforts have been made on training low-bit weight DNNs. In this paper, we focus on training ternary weight {-1, 0, ＋1} networks which can avoid multiplications and dramatically reduce the memory and computation requirements. A ternary weight network can be considered as a sparser version of the binary weight counterpart by replacing some -1s or 1s in the binary weights with 0s, thus leading to more efficient inference but more memory cost. However, the existing approaches to train ternary weight networks cannot control the sparsity (i.e., percentage of 0s) of the ternary weights, which undermines the advantage of ternary weights. In this paper, we propose to our best knowledge the first sparsity-control approach (SCA) to train ternary weight networks, which is simply achieved by a weight discretization regularizer (WDR). SCA is different from all the existing regularizer-based approaches in that it can control the sparsity of the ternary weights through a controller α and does not rely on gradient estimators. We theoretically and empirically show that the sparsity of the trained ternary weights is positively related to α. SCA is extremely simple, easy-to-implement, and is shown to consistently outperform the state-of-the-art approaches significantly over several benchmark datasets and even matches the performances of the full-precision weight counterparts.

Efficient Relational Sentence Ordering Network.

In this paper, we propose a novel deep Efficient Relational Sentence Ordering Network (referred to as ERSON) by leveraging pre-trained language model in both encoder and decoder architectures to strengthen the coherence modeling of the entire model. Specifically, we first introduce a divide-and-fuse BERT (referred to as DF-BERT), a new refactor of BERT network, where lower layers in the improved model encode each sentence in the paragraph independently, which are shared by different sentence pairs, and the higher layers learn the cross-attention between sentence pairs jointly. It enables us to capture the semantic concepts and contextual information between the sentences of the paragraph, while significantly reducing the runtime and memory consumption without sacrificing the model performance. Besides, a Relational Pointer Decoder (referred to as RPD) is developed, which utilizes the pre-trained Next Sentence Prediction (NSP) task of BERT to capture the useful relative ordering information between sentences to enhance the order predictions. In addition, a variety of knowledge distillation based losses are added as auxiliary supervision to further improve the ordering performance. The extensive evaluations on Sentence Ordering, Order Discrimination, and Multi-Document Summarization tasks show the superiority of ERSON to the state-of-the-art ordering methods.

Complementary Calibration: Boosting General Continual Learning with Collaborative Distillation and Self-Supervision.

General Continual Learning (GCL) aims at learning from non independent and identically distributed stream data without catastrophic forgetting of the old tasks that don't rely on task boundaries during both training and testing stages. We reveal that the relation and feature deviations are crucial problems for catastrophic forgetting, in which relation deviation refers to the deficiency of the relationship among all classes in knowledge distillation, and feature deviation refers to indiscriminative feature representations. To this end, we propose a Complementary Calibration (CoCa) framework by mining the complementary model's outputs and features to alleviate the two deviations in the process of GCL. Specifically, we propose a new collaborative distillation approach for addressing the relation deviation. It distills model's outputs by utilizing ensemble dark knowledge of new model's outputs and reserved outputs, which maintains the performance of old tasks as well as balancing the relationship among all classes. Furthermore, we explore a collaborative self-supervision idea to leverage pretext tasks and supervised contrastive learning for addressing the feature deviation problem by learning complete and discriminative features for all classes. Extensive experiments on six popular datasets show that our CoCa framework achieves superior performance against state-of-the-art methods. Code is available at https://github.com/lijincm/CoCa.

Semantic-Guided Class-Imbalance Learning Model for Zero-Shot Image Classification.

In this article, we focus on the task of zero-shot image classification (ZSIC) that equips a learning system with the ability to recognize visual images from unseen classes. In contrast to the traditional image classification, ZSIC more easily suffers from the class-imbalance issue since it is more concerned with the class-level knowledge transferring capability. In the real world, the sample numbers of different categories generally follow a long-tailed distribution, and the discriminative information in the sample-scarce seen classes is hard to transfer to the related unseen classes in the traditional batch-based training manner, which degrades the overall generalization ability a lot. To alleviate the class-imbalance issue in ZSIC, we propose a sample-balanced training process to encourage all training classes to contribute equally to the learned model. Specifically, we randomly select the same number of images from each class across all training classes to form a training batch to ensure that the sample-scarce classes contribute equally as those classes with sufficient samples during each iteration. Considering that the instances from the same class differ in class representativeness, we further develop an efficient semantic-guided feature fusion model to obtain the discriminative class visual prototype for the following visual-semantic interaction process via distributing different weights to the selected samples based on their class representativeness. Extensive experiments on three imbalanced ZSIC benchmark datasets for both traditional ZSIC and generalized ZSIC tasks demonstrate that our approach achieves promising results, especially for the unseen categories that are closely related to the sample-scarce seen categories. Besides, the experimental results on two class-balanced datasets show that the proposed approach also improves the classification performance against the baseline model.

A novel formation pathway of Nε-(carboxyethyl)lysine from lactic acid during high temperature exposure in wheat sourdough bread and chemical model.

Nε-(carboxymethyl)lysine (CML) and Nε-(carboxyethyl)lysine (CEL) have been the most extensively studied advanced glycation end-products (AGEs) in foods. Their formation mechanism, especially the latter, has not been clearly delineated in fermented food. In this work, the relative contents of CEL and CML were evaluated in a sourdough-bread and a silica solid chemical model. Lactic acid (LA) content in the sourdough increased with fermentation time that was accompanied by an increase in CEL, but not CML content in the bread. The role of LA as a precursor for CEL was supported by a positive significant correlation between LA and CEL contents, and further analysis using isotope-labeled LA (LA-13C3) revealed that the three carbon atoms of LA were incorporated into CEL. These findings for the first time indicate LA may function as a precursor to promote CEL formation in sourdough bread that merits further investigation.

Indole-3-propionic acid alleviates ischemic brain injury in a mouse middle cerebral artery occlusion model.

Increasing evidence highlights the importance of gut microbiota and its metabolites as an environmental factor affecting ischemic stroke. However, the role of microbial indole metabolites in ischemic stroke remains largely unknown. Here, we evaluated the effects and the underlying mechanism of indole-3-propionic acid (IPA) in a mouse model of acute middle cerebral artery occlusion (MCAO) and the mechanisms underlying these effects. We collected blood samples and evaluated serum indole derivatives levels using ultra-performance liquid chromatography with tandem mass spectrometry (UPLC-MS) in 8-10-week-old male C57 mice undergoing MCAO or sham. Intragastric IPA administration (400 µg/20 g/d) was performed in mice with MCAO, and its effects and mechanisms were assessed. We found that the serum IPA levels were significantly lower in mice with MCAO than in sham-treated subjects. 16S rRNA gene sequencing revealed that IPA treatment ameliorated the MCAO-induced alterations of the gut microbiome structure, specifically reshaping the microbial community composition in mice with MCAO to resemble that in the mice from the control group, with an increase in the abundance of probiotics and a decrease in the abundance of harmful bacteria. IPA repaired the integrity of the intestinal barrier and regulated the activities of regulatory T cells (Tregs) and Th17 cells in the gut-associated lymphoid tissue. Intragastric IPA administration effectively alleviated neuroinflammation, neurological impairment and brain infarction. Of note, Tregs in the IPA treatment group inhibited A1 reactive astrogliosis in vitro. The beneficial effects of IPA are thus mediated by the gut microbiota, which could enable the development of prebiotics for microbiome-based treatments for ischemic stroke.

Characterization and immunomodulatory effect of an alkali-extracted galactomannan from Morchella esculenta.

In our continuous exploration for bioactive polysaccharides, a novel polysaccharide FMP-2 was isolated and purified from the fruiting bodies of Morchella esculenta by alkali-assisted extraction. FMP-2 had an average molecular weight of 1.09 × 106 Da and contained mannose, glucuronic acid, glucose, galactose, and arabinose in a molar ratio of 4.10:0.22:1.00:5.75:0.44. The backbone of FMP-2 mainly consisted of 1,2-α-D-Galp, 1,6-α-D-Galp, and 1,4-α-D-Manp, with branches of 1,4,6-α-D-Manp and 1,2,6-α-D-Galp. FMP-2 can stimulate phagocytosis and promote the secretion of NO, ROS, and cytokines like IL-6, IL-1ß, and TNF-α in RAW264.7 cells ranging from 25 to 400 µg/mL. FMP-2 had great repairing effect on the immune injury of zebrafish induced by chloramphenicol. The phagocytosis ability of zebrafish macrophages and the proliferation of neutrophils can be greatly enhanced by polysaccharide FMP-2 with concentrations from 50 to 200 µg/mL. These findings suggest that FMP-2 might be used as a potential immunomodulator in the food and pharmaceutical industries.

Efficient Person Search via Expert-Guided Knowledge Distillation.

The person search problem aims to find the target person in the scene images, which presents high demands for both effectiveness and efficiency. In this paper, we present a unified person search framework which jointly handles the two demands for real-world applications. We explore the technique of knowledge distillation (KD), which allows the student network to share capabilities of the deep expert networks with much fewer parameters and less computing time. To achieve this, we describe an efficient person search network and a set of deep and well-engineered expert networks, to build a tiny and compact model that can approximate the representations of the expert networks in a multitask learning manner. We present extensive experiments on three customized student networks with different scales of networks and show strong performance compared to the state-of-the-art methods on both mean average precision and top-1 accuracies. We further demonstrate the efficiency of the proposed network at 120 frames/s in the feedforward time with only a little sacrifice on the accuracy.

Inhibition Mechanism of L-Cysteine on Maillard Reaction by Trapping 5-Hydroxymethylfurfural.

The Maillard reaction (MR) can affect the color, flavor, organoleptic properties, and nutritional value of food. Sometimes, MR is undesirable due to lowering the nutrient utilization, producing harmful neo-formed compounds, etc. In this case, it is necessary to control MR. Some chemical substances, such as phenolic acid, vitamins, aminoguanidine, and thiols extracted from garlic or onion, can effectively prevent MR. In this study, L-cysteine (L-cys) was found to inhibit MR after screening 10 sulfhydryl compounds by comparing their ability to mitigate browning. The inhibition mechanism was speculated to be related to the removal of 5-hydroxymethylfurfural (HMF), a key mid-product of MR. The reaction product of HMF and L-cys was identified and named as 1-dicysteinethioacetal-5-hydroxymethylfurfural (DCH) according to the mass spectrum and nuclear magnetic resonance spectrum of the main product. Furthermore, DCH was detected in the glutamic-fructose mixture after L-cys was added. In addition, the production of DCH also increased with the addition of L-cys. It also was worth noting that DCH showed no cell toxicity to RAW 264.7 cells. Moreover, the in vitro assays indicated that DCH had anti-inflammatory and antioxidant activities. In conclusion, L-cys inhibits MR by converting HMF into another adduct DCH with higher safety and health benefits. L-cys has the potential to be applied as an inhibitor to prevent MR during food processing and storage.

Multitask Non-Autoregressive Model for Human Motion Prediction.

Human motion prediction, which aims at predicting future human skeletons given the past ones, is a typical sequence-to-sequence problem. Therefore, extensive efforts have been devoted to exploring different RNN-based encoder-decoder architectures. However, by generating target poses conditioned on the previously generated ones, these models are prone to bringing issues such as error accumulation problem. In this paper, we argue that such issue is mainly caused by adopting autoregressive manner. Hence, a novel Non-AuToregressive model (NAT) is proposed with a complete non-autoregressive decoding scheme, as well as a context encoder and a positional encoding module. More specifically, the context encoder embeds the given poses from temporal and spatial perspectives. The frame decoder is responsible for predicting each future pose independently. The positional encoding module injects positional signal into the model to indicate the temporal order. Besides, a multitask training paradigm is presented for both low-level human skeleton prediction and high-level human action recognition, resulting in the considerable improvement for the prediction task. Our approach is evaluated on Human3.6M and CMU-Mocap benchmarks and outperforms state-of-the-art autoregressive methods.