Optical neuromorphic processing with soliton crystal Kerr microcombs: Scaling the network in size and speed

Convolutional neural networks (CNNs), inspired by biological visual cortex systems, are a powerful category of artificial neural networks that can extract the hierarchical features of raw data to greatly reduce the network parametric complexity and enhance the predicting accuracy. They are of significant interest for machine learning tasks such as computer vision, speech recognition, playing board games and medical diagnosis [1-7]. Optical neural networks offer the promise of dramatically accelerating computing speed to overcome the inherent bandwidth bottleneck of electronics. Here, we demonstrate a universal optical vector convolutional accelerator operating beyond 10 Tera-OPS (TOPS - operations per second), generating convolutions of images of 250,000 pixels with 8-bit resolution for 10 kernels simultaneously — enough for facial image recognition. We then use the same hardware to sequentially form a deep optical CNN with ten output neurons, achieving successful recognition of full 10 digits with 900 pixel handwritten digit images with 88% accuracy. Our results are based on simultaneously interleaving temporal, wavelength and spatial dimensions enabled by an integrated microcomb source. We show that this approach is scalable and trainable to much more complex networks for demanding applications such as unmanned vehicle and real-time video recognition.Keywords: Optical neural networks, neuromorphic processor, microcomb, convolutional accelerator

Research on Chest Disease Recognition Based on Deep Hierarchical Learning Algorithm

Chest X-ray has become one of the most common ways in diagnostic radiology exams, and this technology assists expert radiologists with finding the patients at potential risk of cardiopathy and lung diseases. However, it is still a challenge for expert radiologists to assess thousands of cases in a short period so that deep learning methods are introduced to tackle this problem. Since the diseases have correlations with each other and have hierarchical features, the traditional classification scheme could not achieve a good performance. In order to extract the correlation features among the diseases, some GCN-based models are introduced to combine the features extracted from the images to make prediction. This scheme can work well with the high quality of image features, so backbone with high computation cost plays a vital role in this scheme. However, a fast prediction in diagnostic radiology is also needed especially in case of emergency or region with low computation facilities, so we proposed an efficient convolutional neural network with GCN, which is named SGGCN, to meet the need of efficient computation and considerable accuracy. SGGCN used SGNet-101 as backbone, which is built by ShuffleGhost Block (Huang et al., 2021) to extract features with a low computation cost. In order to make sufficient usage of the information in GCN, a new GCN architecture is designed to combine information from different layers together in GCNM module so that we can utilize various hierarchical features and meanwhile make the GCN scheme faster. The experiment on CheXPert datasets illustrated that SGGCN achieves a considerable performance. Compared with GCN and ResNet-101 (He et al., 2015) backbone (test AUC 0.8080, parameters 4.7M and FLOPs 16.0B), the SGGCN achieves 0.7831 (−3.08%) test AUC with parameters 1.2M (−73.73%) and FLOPs 3.1B (−80.82%), where GCN with MobileNet (Sandler and Howard, 2018) backbone achieves 0.7531 (−6.79%) test AUC with parameters 0.5M (−88.46%) and FLOPs 0.66B (−95.88%).

Discriminative Siamese Tracker Based on Multi-Channel-Aware and Adaptive Hierarchical Deep Features

Most existing Siamese trackers mainly use a pre-trained convolutional neural network to extract target features. However, due to the weak discrimination of the target and background information of pre-trained depth features, the performance of the Siamese tracker can be significantly degraded when facing similar targets or changes in target appearance. This paper proposes a multi-channel-aware and adaptive hierarchical deep features module to enhance the discriminative ability of the tracker. Firstly, through the multi-channel-aware deep features module, the importance values of feature channels are obtained from both the target details and overall information, to identify more important feature channels. Secondly, by introducing the adaptive hierarchical deep features module, the importance of each feature layer can be determined according to the response value of each frame, so that the hierarchical features can be integrated to represent the target, which can better adapt to changes in the appearance of the target. Finally, the proposed two modules are integrated into the Siamese framework for target tracking. The Siamese network used in this paper is a two-input branch symmetric neural network with two input branches, and they share the same weights, which are widely used in the field of target tracking. Experiments on some Benchmarks show that the proposed Siamese tracker has several points of improvement compared to the baseline tracker.

A study of the labor process from a technology transformation perspective: the case of internet virtual teams

This article presents an empirical study of the labor process of internet virtual teams. It argues that organizations with a “horizontally virtual and vertically real” structure face a dilemma in the virtual team labor process. While a culture of engineers, which embodies equality, liberty, and cooperation, is the cultural basis of the virtual team, management is bureaucratic, emphasizing individual interests and hierarchical features. The coexistence of the two leads to cooperation and division of labor in virtual teams. Essentially, this is a compromising institutional arrangement adopted by corporations to triangulate technology culture and managerial control to obtain surplus value. Based on the preceding discussion, this paper ends by proposing a new theoretical framework for studying the labor process under the technological conditions of the internet.

GaitVision: Real-Time Extraction of Gait Parameters Using Residual Attention Network

Gait walking patterns are one of the key research topics in natural biometrics. The temporal information of the unique gait sequence of a person is preserved and used as a powerful data for access. Often there is a dive into the flexibility of gait sequence due to unstructured and unnecessary sequences that tail off the necessary sequence constraints. The authors in this work present a novel perspective, which extracts useful gait parameters regarded as independent frames and patterns. These patterns and parameters mark as unique signature for each subject in access authentication. This information extracted learns to identify the patterns associated to form a unique gait signature for each person based on their style, foot pressure, angle of walking, angle of bending, acceleration of walk, and step-by-step distance. These parameters form a unique pattern to plot under unique identity for access authorization. This sanitized data of patterns is further passed to a residual deep convolution network that automatically extracts the hierarchical features of gait pattern signatures. The end layer comprises of a Softmax classifier to classify the final prediction of the subject identity. This state-of-the-art work creates a gait-based access authentication that can be used in highly secured premises. This work was specially designed for Defence Department premises authentication. The authors have achieved an accuracy of 90 % ± 1.3 % in real time. This paper mainly focuses on the assessment of the crucial features of gait patterns and analysis of gait patterns research.

Lung cancer subtype classification using histopathological images based on weakly supervised multi-instance learning

Subtype classification plays a guiding role in the clinical diagnosis and treatment of non-small-cell lung cancer (NSCLC). However, due to the gigapixel of whole slide images (WSIs) and the absence of definitive morphological features, most automatic subtype classification methods for NSCLC require manually delineating the regions of interest (ROIs) on WSIs. In this paper, a weakly supervised framework is proposed for accurate subtype classification while freeing pathologists from pixel-level annotation. With respect to the characteristics of histopathological images, we design a two-stage structure with ROI localization and subtype classification. We first develop a method called MR-EM-CNN (multi-resolution expectation-maximization convolutional neural network) to locate ROIs for subsequent subtype classification. The EM algorithm is introduced to select the discriminative image patches for training a patch-wise network, with only WSI-wise labels available. A multi-resolution mechanism is designed for fine localization, similar to the coarse-to-fine process of manual pathological analysis. In the second stage, we build a novel hierarchical attention multi-scale network (HMS) for subtype classification. HMS can capture multi-scale features flexibly driven by the attention module and implement hierarchical features interaction. Experimental results on the 1002-patient Cancer Genome Atlas dataset achieved an AUC of 0.9602 in the ROI localization and an AUC of 0.9671 for subtype classification. The proposed method shows superiority compared with other algorithms in the subtype classification of NSCLC. The proposed framework can also be extended to other classification tasks with WSIs.

VTG-Net: A CNN Based Vessel Topology Graph Network for Retinal Artery/Vein Classification

From diagnosing cardiovascular diseases to analyzing the progression of diabetic retinopathy, accurate retinal artery/vein (A/V) classification is critical. Promising approaches for A/V classification, ranging from conventional graph based methods to recent convolutional neural network (CNN) based models, have been known. However, the inability of traditional graph based methods to utilize deep hierarchical features extracted by CNNs and the limitations of current CNN based methods to incorporate vessel topology information hinder their effectiveness. In this paper, we propose a new CNN based framework, VTG-Net (vessel topology graph network), for retinal A/V classification by incorporating vessel topology information. VTG-Net exploits retinal vessel topology along with CNN features to improve A/V classification accuracy. Specifically, we transform vessel features extracted by CNN in the image domain into a graph representation preserving the vessel topology. Then by exploiting a graph convolutional network (GCN), we enable our model to learn both CNN features and vessel topological features simultaneously. The final predication is attained by fusing the CNN and GCN outputs. Using a publicly available AV-DRIVE dataset and an in-house dataset, we verify the high performance of our VTG-Net for retinal A/V classification over state-of-the-art methods (with ~2% improvement in accuracy on the AV-DRIVE dataset).

Design of Multi-Receptive Field Fusion-Based Network for Surface Defect Inspection on Hot-Rolled Steel Strip Using Lightweight Dataset

With the advancement of industrial intelligence, defect recognition has become an indispensable part of facilitating surface quality in the steel manufacturing process. To assure product quality, most previous studies were typically trained with many defect samples. Nonetheless, a large quantity of defect samples is difficult to obtain, owing to the rare occurrence of defects. In general, deep learning-based methods underperformed as they have inherent limitations due to inadequate information, thereby restraining the application of models. In this study, a two-level Gaussian pyramid is applied to decompose raw data into different resolution levels simultaneously filtering the noises to acquire compact and representative features. Subsequently, a multi-receptive field fusion-based network (MRFFN) is developed to learn the hierarchical features and synthesize the respective prediction scores to form the final recognition result. As a result, the proposed method is capable of exhibiting an outstanding performance of 99.75% when trained using a lightweight dataset. In addition, the experiments conducted using the disturbance defect dataset showed the robustness of the proposed MRFFN against common noises and motion blur.

An Engineering Domain Knowledge-Based Framework for Modelling Highly Incomplete Industrial Data

The missing values in industrial data restrict the applications. Although this incomplete data contains enough information for engineers to support subsequent development, there are still too many missing values for algorithms to establish precise models. This is because the engineering domain knowledge is not considered, and valuable information is not fully captured. Therefore, this article proposes an engineering domain knowledge-based framework for modelling incomplete industrial data. The raw datasets are partitioned and processed at different scales. Firstly, the hierarchical features are combined to decrease the missing ratio. In order to fill the missing values in special data, which is identified for classifying the samples, samples with only part of the features presented are fully utilized instead of being removed to establish local imputation model. Then samples are divided into different groups to transfer the information. A series of industrial data is analyzed for verifying the feasibility of the proposed method.