Molecular dynamics is a powerful simulation tool to explore material properties. Most of the realistic material systems are too large to be simulated with first-principles molecular dynamics. Classical molecular dynamics has lower computational cost but requires accurate force fields to achieve chemical accuracy. In this work, we develop a symmetry-adapted graph neural networks framework, named molecular dynamics graph neural networks (MDGNN), to construct force fields automatically for molecular dynamics simulations for both molecules and crystals. This architecture consistently preserves the translation, rotation and permutation invariance in the simulations. We propose a new feature engineering method including higher order contributions and show that MDGNN accurately reproduces the results of both classical and first-principles molecular dynamics. We also demonstrate that force fields constructed by the model has good transferability. Therefore, MDGNN provides an efficient and promising option for molecular dynamics simulations of large scale systems with high accuracy.
Recently we proposed an all-deep-learning minimum variance distortionless response (ADL-MVDR) method where the unstable matrix inverse and principal component analysis (PCA) operations in the MVDR were replaced by recurrent neural networks (RNNs). However, it is not clear whether the success of the ADL-MVDR is owed to the calculated covariance matrices or following the MVDR formula. In this work, we demonstrate the importance of the calculated covariance matrices and propose three types of generalized RNN beamformers (GRNN-BFs) where the beamforming solution is beyond the MVDR and optimal. The GRNN-BFs could predict the frame-wise beamforming weights by leveraging on the temporal modeling capability of RNNs. The proposed GRNN-BF method obtains better performance than the state-of-the-art ADL-MVDR and the traditional mask-based MVDR methods in terms of speech quality (PESQ), speech-to-noise ratio (SNR), and word error rate (WER).
Many purely neural network based speech separation approaches have been proposed that greatly improve objective assessment scores, but they often introduce nonlinear distortions that are harmful to automatic speech recognition (ASR). Minimum variance distortionless response (MVDR) filters strive to remove nonlinear distortions, however, these approaches either are not optimal for removing residual (linear) noise, or they are unstable when used jointly with neural networks. In this study, we propose a multi-channel multi-frame (MCMF) all deep learning (ADL)-MVDR approach for target speech separation, which extends our preliminary multi-channel ADL-MVDR approach. The MCMF ADL-MVDR handles different numbers of microphone channels in one framework, where it addresses linear and nonlinear distortions. Spatio-temporal cross correlations are also fully utilized in the proposed approach. The proposed system is evaluated using a Mandarin audio-visual corpora and is compared with several state-of-the-art approaches. Experimental results demonstrate the superiority of our proposed framework under different scenarios and across several objective evaluation metrics, including ASR performance.
Recently, deep learning has been widely used in the field of vehicle re-identification. When training a deep model, softmax loss is usually used as a supervision tool. However, the softmax loss performs well for closed-set tasks, but not very well for open-set tasks. In this paper, we sum up five shortcomings of center loss and solved all of them by proposing a dual distance center loss (DDCL). Especially we solve the shortcoming that center loss must combine with the softmax loss to supervise training the model, which provides us with a new perspective to examine the center loss. In addition, we verify the inconsistency between the proposed DDCL and softmax loss in the feature space, which makes the center loss no longer be limited by the softmax loss in the feature space after removing the softmax loss. To be specifically, we add the Pearson distance on the basis of the Euclidean distance to the same center, which makes all features of the same class be confined to the intersection of a hypersphere and a hypercube in the feature space. The proposed Pearson distance strengthens the intra-class compactness of the center loss and enhances the generalization ability of center loss. Moreover, by designing a Euclidean distance threshold between all center pairs, which not only strengthens the inter-class separability of center loss, but also makes the center loss (or DDCL) works well without the combination of softmax loss. We apply DDCL in the field of vehicle re-identification named VeRi-776 dataset and VehicleID dataset. And in order to verify its good generalization ability, we also verify it in two datasets commonly used in the field of person re-identification named MSMT17 dataset and Market1501 dataset.
This paper proposes a new paradigm for handling far-field multi-speaker data in an end-to-end neural network manner, called directional automatic speech recognition (D-ASR), which explicitly models source speaker locations. In D-ASR, the azimuth angle of the sources with respect to the microphone array is defined as a latent variable. This angle controls the quality of separation, which in turn determines the ASR performance. All three functionalities of D-ASR: localization, separation, and recognition are connected as a single differentiable neural network and trained solely based on ASR error minimization objectives. The advantages of D-ASR over existing methods are threefold: (1) it provides explicit speaker locations, (2) it improves the explainability factor, and (3) it achieves better ASR performance as the process is more streamlined. In addition, D-ASR does not require explicit direction of arrival (DOA) supervision like existing data-driven localization models, which makes it more appropriate for realistic data. For the case of two source mixtures, D-ASR achieves an average DOA prediction error of less than three degrees. It also outperforms a strong far-field multi-speaker end-to-end system in both separation quality and ASR performance.
Despite great recent advances in visual tracking, its further development, including both algorithm design and evaluation, is limited due to lack of dedicated large-scale benchmarks. To address this problem, we present LaSOT, a high-quality Large-scale Single Object Tracking benchmark. LaSOT contains a diverse selection of 85 object classes, and offers 1,550 totaling more than 3.87 million frames. Each video frame is carefully and manually annotated with a bounding box. This makes LaSOT, to our knowledge, the largest densely annotated tracking benchmark. Our goal in releasing LaSOT is to provide a dedicated high quality platform for both training and evaluation of trackers. The average video length of LaSOT is around 2,500 frames, where each video contains various challenge factors that exist in real world video footage,such as the targets disappearing and re-appearing. These longer video lengths allow for the assessment of long-term trackers. To take advantage of the close connection between visual appearance and natural language, we provide language specification for each video in LaSOT. We believe such additions will allow for future research to use linguistic features to improve tracking. Two protocols, full-overlap and one-shot, are designated for flexible assessment of trackers. We extensively evaluate 48 baseline trackers on LaSOT with in-depth analysis, and results reveal that there still exists significant room for improvement. The complete benchmark, tracking results as well as analysis are available at http://vision.cs.stonybrook.edu/~lasot/.
Speech enhancement and speech separation are two related tasks, whose purpose is to extract either one or more target speech signals, respectively, from a mixture of sounds generated by several sources. Traditionally, these tasks have been tackled using signal processing and machine learning techniques applied to the available acoustic signals. More recently, visual information from the target speakers, such as lip movements and facial expressions, has been introduced to speech enhancement and speech separation systems, because the visual aspect of speech is essentially unaffected by the acoustic environment. In order to efficiently fuse acoustic and visual information, researchers have exploited the flexibility of data-driven approaches, specifically deep learning, achieving state-of-the-art performance. The ceaseless proposal of a large number of techniques to extract features and fuse multimodal information has highlighted the need for an overview that comprehensively describes and discusses audio-visual speech enhancement and separation based on deep learning. In this paper, we provide a systematic survey of this research topic, focusing on the main elements that characterise the systems in the literature: visual features; acoustic features; deep learning methods; fusion techniques; training targets and objective functions. We also survey commonly employed audio-visual speech datasets, given their central role in the development of data-driven approaches, and evaluation methods, because they are generally used to compare different systems and determine their performance. In addition, we review deep-learning-based methods for speech reconstruction from silent videos and audio-visual sound source separation for non-speech signals, since these methods can be more or less directly applied to audio-visual speech enhancement and separation.
Face detection from low-light images is challenging due to limited photos and inevitable noise, which, to make the task even harder, are often spatially unevenly distributed. A natural solution is to borrow the idea from multi-exposure, which captures multiple shots to obtain well-exposed images under challenging conditions. High-quality implementation/approximation of multi-exposure from a single image is however nontrivial. Fortunately, as shown in this paper, neither is such high-quality necessary since our task is face detection rather than image enhancement. Specifically, we propose a novel Recurrent Exposure Generation (REG) module and couple it seamlessly with a Multi-Exposure Detection (MED) module, and thus significantly improve face detection performance by effectively inhibiting non-uniform illumination and noise issues. REG produces progressively and efficiently intermediate images corresponding to various exposure settings, and such pseudo-exposures are then fused by MED to detect faces across different lighting conditions. The proposed method, named REGDet, is the first `detection-with-enhancement' framework for low-light face detection. It not only encourages rich interaction and feature fusion across different illumination levels, but also enables effective end-to-end learning of the REG component to be better tailored for face detection. Moreover, as clearly shown in our experiments, REG can be flexibly coupled with different face detectors without extra low/normal-light image pairs for training. We tested REGDet on the DARK FACE low-light face benchmark with thorough ablation study, where REGDet outperforms previous state-of-the-arts by a significant margin, with only negligible extra parameters.
Deep convolutional neural networks (CNNs) with strong expressive ability have achieved impressive performances on single image super-resolution (SISR). However, their excessive amounts of convolutions and parameters usually consume high computational cost and more memory storage for training a SR model, which limits their applications to SR with resource-constrained devices in real world. To resolve these problems, we propose a lightweight enhanced SR CNN (LESRCN-N) with three successive sub-blocks, an information extraction and enhancement block (IEEB), a reconstruction block (RB) and an information refinement block (IRB). Specifically, the IEEB extracts hierarchical low-resolution (LR) features and aggregates the obtained features step-by-step to increase the memory ability of the shallow layers on deep layers for SISR. To remove redundant information obtained, a heterogeneous architecture is adopted in the IEEB. After that, the RB converts low-frequency features into high-frequency features by fusing global and local features, which is complementary with the IEEB in tackling the long-term dependency problem. Finally, the IRB uses coarse high-frequency features from the RB to learn more accurate SR features and construct a SR image. The proposed LESRCNN can obtain a high-quality image by a model for different scales. Extensive experiments demonstrate that the proposed LESRCNN outperforms state-of-the-arts on SISR in terms of qualitative and quantitative evaluation.