Search

The goal of the work described here was to construct a deep learning–based intelligent diagnostic model for ophthalmic ultrasound images to provide auxiliary analysis for the intelligent clinical diagnosis of posterior ocular segment diseases. The InceptionV3–Xception fusion model was established by using two pre-trained network models—InceptionV3 and Xception—in series to achieve multilevel feature extraction and fusion, and a classifier more suitable for the multiclassification recognition task of ophthalmic ultrasound images was designed to classify 3402 ophthalmic ultrasound images. The accuracy, macro-average precision, macro-average sensitivity, macro-average F1 value, subject working feature curves and area under the curve were used as model evaluation metrics, and the credibility of the model was assessed by testing the decision basis of the model using a gradient-weighted class activation mapping method. The accuracy, precision, sensitivity and area under the subject working feature curve of the InceptionV3–Xception fusion model on the test set reached 0.9673, 0.9521, 0.9528 and 0.9988, respectively. The model decision basis was consistent with the clinical diagnosis basis of the ophthalmologist, which proves that the model has good reliability. The deep learning–based ophthalmic ultrasound image intelligent diagnosis model can accurately screen and identify five posterior ocular segment diseases, which is beneficial to the intelligent development of ophthalmic clinical diagnosis. [ABSTRACT FROM AUTHOR]

Orbital angular momentum (OAM) multiplexing technology is a highly promising approach that can significantly increase the data capacity of optical communication systems. However, traditional optical communication systems are prone to atmospheric disturbances such as turbulence, which can degrade transmission quality. To mitigate this issue, multimode fibers (MMFs) have been introduced to reduce the impact of turbulence on signals. Moreover, previous studies have primarily focused on the identification of single vortex beams, facing challenges in accurately recognizing multiplexed modes. To address this challenge, this chapter proposes a multi-label image classification optimization algorithm based on transfer learning. By utilizing the pre-trained MobileNet V2 model as a feature extractor, this network structure can accurately identify 8-bit, 16-bit, and 24-bit multiplexed OAM from speckle patterns in multimode fibers, even with small sample datasets, achieving classification accuracies exceeding 95%. This method overcomes the limitations of traditional optical communication systems that are susceptible to atmospheric disturbances, providing new possibilities for long-distance transmission and increased data capacity in optical communication systems. • We use MMF to transmit vortex light, which significantly reduces the impact of atmospheric turbulence. • A multi-label image classification optimization algorithm based on transfer learning, ML-OAM-DemuxNet, is proposed to identify speckle patterns. • Cutting the center part of the speckle pattern for recognition can greatly reduce the computing cost of the network while maintaining the accuracy [ABSTRACT FROM AUTHOR]

The joint torque estimation model based on surface electromyography (sEMG) signals needs to recalibrate the model parameters when used for joint torque estimation at different volunteers. However, the feature transferability drops significantly in higher task-specific layers, which reduce the performance of the model. The generalization of joint torque estimation models among different volunteers has become a challenge. In this paper, a fast calibration method based on a deep adaptive regression network is proposed to reduce the frequency and time of model recalibration when volunteers change. The proposed method associates source domain (SD) data with target domain (TD) data through migration learning. The maximum mean difference (MMD) adaptive method is used to migrate the model trained in the SD dataset to the TD dataset, which avoids the recalibration required in the joint torque estimation model. The experimental results show that the proposed method can improve the migration ability of the model used for estimating joint torque and achieve the generalization application of the model on different volunteer datasets. Moreover, the proposed method can reduce the calibration operations and iterations of the model caused by volunteer changes and improve the speed of joint torque estimation. The optimal normalized root mean square error (NRMSE) and Pearson correlation coefficient (ρ) of knee extension/flexion joint torque estimation reached 0.02198/0.02565 and 0.9892/0.9839, respectively. [ABSTRACT FROM AUTHOR]

The current most advanced CNN-based detection models are nearly not deployable on mobile devices with limited arithmetic power due to problems such as too many redundant parameters and excessive arithmetic power required, and knowledge distillation as a potentially practical model compression approach can alleviate this limitation. In the past, feature-based knowledge distillation algorithms focused more on transferring the local features customized by people and reduced the full grasp of global information in images. To address the shortcomings of traditional feature distillation algorithms, we first improve GAMAttention to learn the global feature representation in images, and the improved attention mechanism can minimize the information loss caused by processing features. Secondly, feature shifting no longer defines manually which features should be shifted, a more interpretable approach is proposed where the student network learns to emulate the high-response feature regions predicted by the teacher network, which increases the end-to-end properties of the model, and feature shifting allows the student network to simulate the teacher network in generating semantically strong feature maps to improve the detection performance of the small model. To avoid learning too many noisy features when learning background features, these two parts of feature distillation are assigned different weights. Finally, logical distillation is performed on the prediction heads of the student and teacher networks. In this experiment, we chose Yolov5 as the base network structure for teacher–student pairs. We improved Yolov5s through attention and knowledge distillation, ultimately achieving a 1.3% performance gain on VOC and a 1.8% performance gain on KITTI. • We use knowledge distillation and attention for yolov5s to improve performance. • We propose the GAMF module that captures both global and local features. • We propose the RFT and join SSIM to shift response features and background features. [ABSTRACT FROM AUTHOR]

• A hybrid SIFC method is proposed to determine the unknown state or new load. • The isolated forest method is introduced to improve the judgment of unknown loads. • A shared parameter migration method is proposed to realize model self-updating. With the development of smart grid technology, optimization of energy structure, improvement of power efficiency and reduction of power consumption have become the major development trends. As a key process of power structure analysis, load identification is gaining increasing attention in smart grids. Although lots of load identification methods have been proposed, the introduction of unknown states or new load types is still a challenge for precise identification. In this paper, a similarity calculation method of the space convex hull overlap rate is proposed to deal with the identification of unknown loads based on the low-dimensional feature space model of Siamese neural networks. Moreover, transfer learning is introduced to realize the category-added learning of load pre-identification model for unknown load. The ultimate aim was to establish the identification and rapid modeling of unknown loads. The public datasets Plaid is used to verify the performance of the proposed method. As a result, the high accuracy of load identification for unknown load is proved. [ABSTRACT FROM AUTHOR]