Search

Emotion recognition plays a crucial role in human-computer interaction, and electroencephalography (EEG) is advantageous for reflecting human emotional states. In this study, we propose MACTN, a hierarchical hybrid model for jointly modeling local and global temporal information. The model is inspired by neuroscience research on the temporal dynamics of emotions. MACTN extracts local emotional features through a convolutional neural network (CNN) and integrates sparse global emotional features through a transformer. Moreover, we employ channel attention mechanisms to identify the most task-relevant channels. Through extensive experimentation on two publicly available datasets, namely THU-EP and DEAP, our proposed method, MACTN, consistently achieves superior classification accuracy and F1 scores compared to other existing methods in most experimental settings. Furthermore, ablation studies have shown that the integration of both self-attention mechanisms and channel attention mechanisms leads to improved classification performance. Finally, an earlier version of this method, which shares the same ideas, won the Emotional BCI Competition's final championship in the 2022 World Robot Contest.

Serverless computing is a popular cloud computing paradigm, which requires low response latency to handle on-demand user requests. There are two prominent techniques employed for reducing the response latency: keep fully initialized containers alive (Warm Container) or reduce the new container startup (cold start) latency. This paper presents the 3rd container startup mode: Hibernate Container, which starts faster than the cold start container mode and consumes less memory than the Warm Container mode. Hibernate Container is essentially a "deflated" Warm Container. Its application memory is swapped out to disk, the freed memory is reclaimed and file based mmap memory is cleaned-up. The Hibernate Container's deflated memory is inflated in response to user requests. As Hibernate Container's application is fully initialized, its response latency is less than the cold start mode; and as the application memory is deflated, its memory consumption is less than the Warm Container mode. Additionally, when a Hibernate Container is "woken up" to process a request, the Woken-up Container has similar response latency to Warm Container but less memory consumption because not all the deflated memory needs to be inflated. We implemented the Hibernate technique as part of the open source Quark secure container runtime project and our test demonstrated that Hibernate Container consumes about 7\% to 25\% of the Warm Container memory. All of this results in a higher deployment density, lower latency and appreciable improvements in the overall system performance.

Cloud-native containerized applications constantly seek high-performance and easy-to-operate container network solutions. RDMA network is a potential enabler with higher throughput and lower latency than the standard TCP/IP network stack. However, several challenges remain in equipping containerized applications with RDMA network: 1) How to deliver transparent improvements without modifying application code; 2) How to integrate RDMA-based network solutions with container orchestration systems; 3) How to efficiently utilize RDMA for container networks. In this paper, we present an RDMA-based container network solution, TCP Socket over RDMA (TSoR), which addresses all the above challenges. To transparently accelerate applications using POSIX socket interfaces without modifications, we integrate TSoR with a container runtime that can intercept system calls for socket interfaces. To be compatible with orchestration systems like Kubernetes, TSoR implements a container network following the Kubernetes network model and satisfies all requirements of the model. To leverage RDMA benefits, TSoR designs a high-performance network stack that efficiently transfers TCP traffic using RDMA network. Thus, TSoR provides a turn-key solution for existing Kubernetes clusters to adopt the high-performance RDMA network with minimal effort. Our evaluation results show that TSoR provides up to 2.3x higher throughput and 64\% lower latency for existing containerized applications, such as Redis key-value store and Node.js web server, with no code changes. TSoR code will be open-sourced.

A transition-metal-free method for the construction of 3-substituted or 3,4-disubstituted quinolines from readily available N,N-dimethyl enaminones and o-aminobenzyl alcohols is reported. The direct oxidative cyclocondensation reaction tolerates broad functional groups, allowing the efficient synthesis of various quinolines in moderate to excellent yields. The reaction involves a C (sp3)-O bond cleavage and a C=N bind and a C=C bond formation during the oxidative cyclization process, and the mechanism was proposed.

Spinal cord injury (SCI) results from various pathogenic factors that destroy the normal structure and function of the spinal cord, subsequently causing sensory, motor, and autonomic nerve dysfunction. SCI is one of the most common causes of disability and death globally. It leads to severe physical and mental injury to patients and causes a substantial economic burden on families and the society. The pathological changes and underlying mechanisms within SCI involve oxidative stress, apoptosis, inflammation, etc. As a traditional therapy, acupuncture has a positive effect promoting the recovery of SCI. Acupuncture-induced neuroprotection includes several mechanisms such as reducing oxidative stress, inhibiting the inflammatory response and neuronal apoptosis, alleviating glial scar formation, promoting neural stem cell differentiation, and improving microcirculation within the injured area. Therefore, the recent studies exploring the mechanism of acupuncture therapy in SCI will help provide a theoretical basis for applying acupuncture and seeking a better treatment target and acupuncture approach for SCI patients.

The aim of this work is to simulate the formation of coaxial electrohydrodynamic jet (CE-Jet) and obtain the optimized working parameters for coaxial electrohydrodynamic focused jet printing (CEFJP), then further direct writing nanoscale structures. In this paper, a pioneering simulation of CE-Jet process based on three phase flow of liquid–liquid–air model was developed. The photoresist (AZ703)/silicone oil was employed to verify the simulation. The simulated result of the CE-Jet consisted with the experimental result very well, which proves the correction of the CE-Jet model. Based on the simulation, the effect of working parameters including the applied voltage and the flow rate of inner coaxial liquid on the morphology and size of the CE-Jet were examined. With the use of the optimized working parameters obtained from the simulation and taking the materials of lead zirconate titanate (PZT) and photoresist (AZ703) as inner liquid, the stable nanoscale inner jet was focused by the compound effect of electrical shearing force, viscous shearing force and internal pressure. Then, the nanostructures of various patterns with diameter of 70 nm were directly printed. The CE-Jet model developed in this work provides a powerful tool to investigate the CEFJP process, and the CEFJP technique exhibited a promising method for producing nanoscale structures for M/NEMS devices.

Both the Dictionary Learning (DL) and Convolutional Neural Networks (CNN) are powerful image representation learning systems based on different mechanisms and principles, however whether we can seamlessly integrate them to improve the per-formance is noteworthy exploring. To address this issue, we propose a novel generalized end-to-end representation learning architecture, dubbed Convolutional Dictionary Pair Learning Network (CDPL-Net) in this paper, which integrates the learning schemes of the CNN and dictionary pair learning into a unified framework. Generally, the architecture of CDPL-Net includes two convolutional/pooling layers and two dictionary pair learn-ing (DPL) layers in the representation learning module. Besides, it uses two fully-connected layers as the multi-layer perception layer in the nonlinear classification module. In particular, the DPL layer can jointly formulate the discriminative synthesis and analysis representations driven by minimizing the batch based reconstruction error over the flatted feature maps from the convolution/pooling layer. Moreover, DPL layer uses l1-norm on the analysis dictionary so that sparse representation can be delivered, and the embedding process will also be robust to noise. To speed up the training process of DPL layer, the efficient stochastic gradient descent is used. Extensive simulations on real databases show that our CDPL-Net can deliver enhanced performance over other state-of-the-art methods., Accepted by the 24th European Conference on Artificial Intelligence (ECAI 2020)

Purpose This paper aims to effectively achieve endmembers and relative abundances simultaneously in hyperspectral image unmixing yield. Hyperspectral unmixing, which is an important step before image classification and recognition, is a challenging issue because of the limited resolution of image sensors and the complex diversity of nature. Unmixing can be performed using different methods, such as blind source separation and semi-supervised spectral unmixing. However, these methods have disadvantages such as inaccurate results or the need for the spectral library to be known a priori. Design/methodology/approach This paper proposes a novel method for hyperspectral unmixing called fuzzy c-means unmixing, which achieves endmembers and relative abundance through repeated iteration analysis at the same time. Findings Experimental results demonstrate that the proposed method can effectively implement hyperspectral unmixing with high accuracy. Originality/value The proposed method present an effective framework for the challenging field of hyperspectral image unmixing.

This dissertation deals with Corporate Social Responsibility (CSR) issues, especially greenwashing (i.e., a firm’s use of information to appear more socially responsible than it actually is). It includes a literature review, an empirical study, and a case study, organized as three chapters. Chapter 1 is a comprehensive review of the literature. It first discusses what CSR is. Then it explores how CSR enhances firm value from the perspectives of different participants in CSR, including managers, consumers, investors, regulators and rating agencies. It also summarizes related theories on CSR, including legitimacy theory, shift of focus theory, resources-based perspective, stakeholder theory. Finally it reviews related theories and previous studies on greenwashing. Chapter 2 discusses the phenomenon of CSR report restatement. Many firms voluntarily issue CSR reports and restate them subsequently. The magnitude of restated changes is generally very small. Based on signaling theory, this study argues that CSR report restatement is a way for firms to signal that they are truly committed to CSR and not simply greenwashing. Thus, the symbolic meaning of a restatement is more important than its actual influence. CSR restatement is a voluntary choice of management rather than a forced choice due to outside pressure. Compliance with the Global Reporting Initiative (GRI), the reporting experience, previous CSR assurance, and membership in a sensitive industry are found to be determinants of CSR restatement. Chapter 3 is a case study of Chinese greenwashing firms. Previous empirical studies provide mixed results on the market reaction to greenwashing firms. This paper introduces the role of media to explain the motivation behind different market reactions. Applying the yearly list of greenwashing firms from South Weekend in China, this case study finds that the media not only disseminates secondary information, but also generates valuable information by issuing lists of firms that engage in greenwashing. The market reaction to greenwashing is associated with the extent of a firm’s greenwashing and whether that firm is on the list for the first time. Furthermore, the media is useful in detecting greenwashing, but fails to monitor whether greenwashing firms improve their corporate social performance in the future.

Herbivorous insects encounter chemical defense in the form of numerous secondary metabolites in host plants.Understanding insect counter-defense mechanisms is important.In this study fifth instar Spodoptera litura insects were exposed to diets containing 0.01% cinnamic acid,0.01% salicylic acid,0.01% xanthotoxin,0.02% quercetin,0.05% flavone and 0.1% coumarin for 48 h.Activities of detoxification enzymes including glutathione S-transferase(GSTs),caboxylesterase(CarE),cytochrome P450(P450) and acetylcholinesterase(AChE) were then determined in midguts and fatbodies,and transcript levels of CYP4M14 and CYP4S9 were examined by semi-quantitative RT-PCR.CarE activity in midguts was increased 1.67-and 1.37-fold after consumption of cinnamic acids and coumarin,respectively.All six plant secondary metabolites induced GSTs activity in the fatbodies.Total P450 activity in the fatbodies was increased 2.93-and 14.50-fold after consumption of cinnamic acid and coumarin,respectively.Consumption of cinnamic acid,xanthotoxin,quercetin and coumarin increased AChE activity 1.53-,1.80-,2.36-and 1.56-fold,respectively.All six plant secondary metabolites induced transcripts of CYP4M14 in the fatbodies of S.litura.Cinnamic acid,quercetin and coumarin strongly induced transcripts of CYP4S9 in the fatbodies.These results indicated that S.litura is able to induce its detoxification enzymes to minimize toxicity from plant secondary metabolites.

In the field of multiple-camera video surveillance, object tracking is attracting more and more attention. Problems such as objects' abrupt motion, occlusion and complex target structures make this field full of challenges. In the paper, a method based on particle filter and sparse representation for large-scale object tracking is proposed. At first, the features of target objects are trained, then we detect the motion region in the high resolution video, using human crowd segmentation algorithm to separate person from the crowd. After getting the region of single person, the features of the region such as color histogram and hash code would be extracted to match with trained features of target objects. According to the performance of feature matching, we find the true targeted object and its smallest rectangle area. In tracking process, discriminative Sparse Similarity Map (SSM) is used to guarantee a good performance of target tracking. Experiment results demonstrate our method can provide high accuracy and robustness.