Search

This paper discusses the reachable set estimation problem of neural networks with mixed delays. Firstly, by means of the maximal Lyapunov–Krasovskii functional, we obtain a non-ellipsoid form of the reachable set. Further more, when calculating the derivative of the maximum Lyapunov functional, the lower bound lemma and reciprocally convex approach method are used to solve the reciprocally convex combination term, which reduce the related decision variables. Secondly, we extend the results to polytopic uncertainties neural networks and consider the case of uncertain differentiable parameters. Finally, two numerical examples and one application example are listed to show the validity of our methods.

Body language is one of the most common ways of expressing human emotion. In this article, we make the first attempt to generate an action video with a specific emotion from a single person image. The goal of the emotion-based action generation task (EBAG) is to generate action videos expressing a specific type of emotion given a single reference image with a full human body. We divide the task into two parts and propose a two-stage framework to generate action videos with specified emotions. At the first stage, we propose an emotion-based pose sequence generation approach (EPOSE-GAN) for translating the emotion to a pose sequence. At the second stage, we generate the target video frames according to the three inputs including the source pose and the target pose as the motion information and the source image as the appearance reference by using conditional GAN model with an online training strategy. Our framework produces the pose sequence and transforms the action independently, which highlights the fundamental role that the high-level pose feature plays in generating action video with a specific emotion. The proposed method has been evaluated on the “Soul Dancer” dataset which is built for action emotion analysis and generation. The experimental results demonstrate that our framework can effectively solve the emotion-based action generation task. However, the gap in the details of the appearance between the generated action video and the real-world video still exists, which indicates that the emotion-based action generation task has great research potential.

The interaction between pyridoxal hydrochloride (HQ) and apoCopC was investigated using Fourier transform infrared spectroscopy (FTIR), isothermal titration calorimetry (ITC), circular dichroism (CD), fluorescence spectroscopy, three-dimensional (3D) fluorescence spectroscopy, fluorescence lifetime, TNS fluorescence and docking methods. FTIR, CD, TNS fluorescence and fluorescence lifetime experiments suggested that the apoCopC conformation was altered by HQ with an increase in the random coil content and a reduction in the β-sheet content. In addition, the data from fluorescence spectroscopy, 3D fluorescence spectroscopy and molecular docking revealed that the binding site of HQ was located in the hydrophobic area of apoCopC, and a redshift of the HQ fluorescence spectra was observed. Furthermore, ITC and fluorescence quenching data manifested that the binding ratio of HQ and apoCopC was 1:1, and the forming constant was calculated to be (7.06 ± 0.21) × 105 M−1. The thermodynamic parameters ΔH and ΔS suggested that the formation of a CopC-HQ complex depended on the hydrophobic force. Furthermore, the average binding distance between tryptophan in apoCopC and HQ was determined by means of Forster non-radioactive resonance energy transfer and molecular docking. The results agreed well with each other. As a redox switch in the modulation of copper, the interaction of apoCopC with small molecules will affect the action of the redox switch. These findings could provide useful information to illustrate the copper regulation mechanism.

We introduce a principled approach for synthesizing new views of a scene given a single source image. Previous methods for novel view synthesis can be divided into image-based rendering methods (e.g., flow prediction) or pixel generation methods. Flow predictions enable the target view to re-use pixels directly, but can easily lead to distorted results. Directly regressing pixels can produce structurally consistent results but generally suffer from the lack of low-level details. In this paper, we utilize an encoder–decoder architecture to regress pixels of a target view. In order to maintain details, we couple the decoder aligned feature maps with skip connections, where the alignment is guided by predicted depth map of the target view. Our experimental results show that our method does not suffer from distortions and successfully preserves texture details with aligned skip connections.

We propose a method for fusing stereo disparity estimation with movement-induced prior information. Instead of independent inference frame-by-frame, we formulate the problem as a non-parametric learning task in terms of a temporal Gaussian process prior with a movement-driven kernel for inter-frame reasoning. We present a hierarchy of three Gaussian process kernels depending on the availability of motion information, where our main focus is on a new gyroscope-driven kernel for handheld devices with low-quality MEMS sensors, thus also relaxing the requirement of having full 6D camera poses available. We show how our method can be combined with two state-of-the-art deep stereo methods. The method either work in a plug-and-play fashion with pre-trained deep stereo networks, or further improved by jointly training the kernels together with encoder-decoder architectures, leading to consistent improvement.

The ability of a flexible pressure sensor to possess zero power consumption in standby mode, high sensitivity, and wide linear-response range is critical in real flexible matrix-based scenes. However, when the conventional flexible pressure sensors are attached on a curved surface, a pseudosignal response is generated because of the normal stress, resulting in a short linear-response range. Here, a flexible piezoresistive pressure sensor with high performance, zero standby power consumption is demonstrated. The flexible pressure sensor is fabricated from polydimethylsiloxane (PDMS)/carbon black (CB), patterned polyimide (PI) spacer layer, and laser-induced graphene (LIG) interdigital electrodes. Benefiting from the hierarchical structure and sufficient roughness of PDMS/CB and LIG interdigital electrodes, the proposed pressure sensors (PDMS/CB/PI/LIG) exhibit high sensitivity (43 kPa-1), large linear-response range (0.4-13.6 kPa), fast response (1800 cycles). The resulting pressure sensor also features zero standby power consumption merit under certain bending conditions (bending angle: 0-5o). Furthermore, the effect of the hole diameter of the PI spacer layer on the performance of the pressure sensors is experimentally and theoretically investigated. As a proof of concept, a bioinspired artificial haptic neuron system has been successfully equipped to modulate the number of lit LED lights. The proposed high-performance pressure sensor has promising potential to be used in flexible and wearable electronics, especially for the applications in actual flexible matrix-based scenes. Ministry of Education (MOE) This work was partially supported by the National Key R&D Program of China (2018YFB1500200), Shenzhen Basic Research Grant: JCYJ20180507182431967, JCYJ20170413153246713, Shenzhen Peacock Technology Innovation Project: KQJSCX20170731165602155, the National Nature Science Foundation of China (11804354, 61574157, 61774164). The authors are also grateful for the support of Singapore Ministry of Education Academic Research Fund Tier 2 (MOE2015-T2-2-010), and Singapore Ministry of Education Academic Research Fund Tier 1 (MOE2019-T1-001-103).

We propose a novel idea for depth estimation from multi-view image-pose pairs, where the model has capability to leverage information from previous latent-space encodings of the scene. This model uses pairs of images and poses, which are passed through an encoder--decoder model for disparity estimation. The novelty lies in soft-constraining the bottleneck layer by a nonparametric Gaussian process prior. We propose a pose-kernel structure that encourages similar poses to have resembling latent spaces. The flexibility of the Gaussian process (GP) prior provides adapting memory for fusing information from previous views. We train the encoder--decoder and the GP hyperparameters jointly end-to-end. In addition to a batch method, we derive a lightweight estimation scheme that circumvents standard pitfalls in scaling Gaussian process inference, and demonstrate how our scheme can run in real-time on smart devices., ICCV 2019

This paper presents a novel method, MaskMVS, to solve depth estimation for unstructured multi-view image-pose pairs. In the plane-sweep procedure, the depth planes are sampled by histogram matching that ensures covering the depth range of interest. Unlike other plane-sweep methods, we do not rely on a cost metric to explicitly build the cost volume, but instead infer a multiplane mask representation which regularizes the learning. Compared to many previous approaches, we show that our method is lightweight and generalizes well without requiring excessive training. We outperform the current state-of-the-art and show results on the sun3d, scenes11, MVS, and RGBD test data sets.

We propose a novel electrochemical method to one-pot synthesize carbon dots/titanium dioxide (CDs/TiO 2 ) nanocomposites for the first time. The as-prepared CDs/TiO 2 nanocomposites exhibit excellent visible photocatalytic activity for degradation of methylene blue (MB) under a 50 W halogen tungsten lamp irradiation.

The recent advancement of flexible pressure sensors leads to promising applications in electronic skins for human-machine interaction and bionic device for intelligent wearable field. The abilities of such sensors to possess zero power consumption in standby mode, high sensitivity and wide linear response range are critical yet highly desirable in real flexible-matrix-based scenes. Despite the myriad of the existing approaches, it is still challenging to fulfill these requirements simultaneously in a single pressure sensor. Herein, we demonstrate a high-performance, zero-standby-power-consumption flexible pressure sensor by introducing a tunable photoresist spacer (PS) sandwiched between the top truncated-pyramid microstructure polydimethylsiloxane (PDMS)/aluminum-doped zinc oxide (AZO) and the bottom polyimide (PI)/gold (Au) interdigital electrode (PDMS-AZO/PS/PI-Au). The fabricated pressure sensor is equipped with bending-insensitive and zero-power-consumption merit under the bending angle of less than 20.5° benefiting from the insulation effect of PS. As a result, the fabricated sensor exhibits ultrahigh sensitivity of 2200 kPa−1 in the ultrawide linear response range of 62 Pa-9.6 kPa and fast response and recovery time (

Context information plays an important role in object detection. DeepID concatenates the global context for classification, while Yolo, SSD, and Crafting use local context information for detection. In this paper, we propose a straightforward method to plug the global context information into the Faster RCNN Framework, namely the Skip Context Connection (SCC). We use SCC to inject the global context into the object representation which skips the RoiPooling layer rather than drops it. Therefore, it can not only leverage the context information but also keep the location accuracy from the RCNN framework. We proposed three principles to construct the SCC blocks: effectiveness means fewer parameters, additivity means the features possess the same meaning, and selectable means soft gated addition. We also evaluate several different SCC blocks. The Gated Additive SCC(GA-SCC) which satisfy the three principles get the best performance. Our experiment results on PASCAL VOC 2007 show that GA-SCC can get the steady 1% improvement over the traditional RCNN method.

In order to strengthen the framework structure of distribution network, promote the agility of its control methods, and improve its intelligent and automation level, Soft Normally-Open Point (SNOP) as a novel power electronic device is installed in distribution network, which can provide accurate control of real and reactive power flow. This paper proposed the dynamic mathematical model of three-port SNOP in dq coordinates, which is able to connect more feeders or AC systems and improve its extensibility and flexibility. The dual closed-loop cooperative control system was adopted to achieve the multiple target control. In order to get the dynamic response characteristics of three-port SNOP, the simulation model was built and tested in the EMTDC/PSCAD platform with different control strategies. The modified IEEE 33-bus power system was used to demonstrate the benefits of SNOP. The simulation results showed that the three-port SNOP with PV control mode and PV control mode have different degree of advantages in balancing feeders load, reducing power loss and improving the node voltage level.

Sulfur, nitrogen co-doped carbon dots (S, N co-doped C-dots) as highly selective fluorescent probe for uric acid (UA) detection were designed. The S, N co-doped C-dots with high quantum yield of 73.1% were prepared by hydrothermal method. It was found that the fluorescence of S, N co-doped C-dots was quenched apparently by hydroxyl radicals from Fenton reaction between H2O2 and Fe(2+). The production of H2O2 originated from the oxidization of UA by uricase. Therefore, an optical biosensor was developed for the detection of UA based on Fenton reaction and enzymatic reaction. Under the optimized conditions, two linear relationships between the ratio of fluorescence quenching of the C-dots and UA concentration were found in the range of 0.08-10µM and 10-50µM, respectively. The detection limit was down to 0.07µM. Moreover, the proposed biosensor was successfully applied to the detection of uric acid in human serum samples.

The graphitic carbon nitride nanodots (g-C3N4-dots) were synthesized by a simple electrochemical "tailoring" process from bulk graphitic carbon nitride (g-C3N4) under alkaline solution for the first time. Compared with the bulk g-C3N4, the novel g-C3N4-dots not only exhibit enhanced fluorescence and excellent dispersion stability in water but also show the reducibility for the reduction of Ag+ to AgNPs at 60°C. The biothiols can bound with Ag+ through formation of biothiol-Ag+ complex to consume the Ag+ and act as capping agent to prevent the growth of AgNPs, which cause the decrease of the absorption peak of the AgNPs. Therefore, an optical sensor was developed for the detection of biothiols based on the change of the plasmon resonance absorption peak of the AgNPs. The proposed method exhibits excellent sensitivity and selectivity to biothiols with low detection limit for cysteine (Cys), homocysteine (Hcy), and glutathione (GSH) with 11.5, 16.1, and 15.5nM, respectively. This method also has been successfully applied for the detection of biothiols in human serum with satisfactory results.