Robot target recognition using deep federated learning

Abstract

Robot target recognition is a critical and fundamental machine vision task. In this paper, InVision, a robot target recognition approach is proposed using deep federated learning. Particularly, deep geometric learning is developed to improve the perception capabilities of convolutional neural networks, and promote the representation maps' resolutions while achieving good recognition performance. Moreover, federated metric learning is constructed to protect user data privacy across multiple devices and relieve the problem of inadequate available labeled training data. To improve the speed of the recognition system, a lightweight deep neural network is presented. Extensive experiments are performed, showing that InVision significantly outperforms the outstanding comparison approaches.

Bayesian‐MLE signal detection for multi‐antenna ambient backscatter communication

“…To further improve the detection performance, multi-task joint learning and other global optimization schemes will be considered in multi-antenna AmBC signal detection. Furthermore, deep learning (DL) has achieved great successes in exploring essential features and uncertainty information [30][31][32]; it may be an excellent solution to solve the problem of AmBC signal detection. In addition, to enhance the detector's robustness, more real-world issues will be explored, such as data with low SNR and time series or frequency information.…”

Feature selection and feature learning in machine learning applications for gas turbines: A review

“…AI highlights the ideology that machines can simulate human intelligence and automatically perform decision-making with appropriate knowledge (Konar, 2018;Xue and Tong, 2019a). ML, a branch of AI, includes many algorithms that automatically extract patterns from datasets, discover correlations between inputs and outputs, and perform designated tasks such as fault classification and anomaly detection (Xue et al, 2021;Xue and Tong, 2019b). Deep learning (DL) is a type of ML that employs the structure of artificial neural networks (ANNs) with multiple hidden layers.…”

Graph convolutional network‐based deep feature learning for cardiovascular disease recognition from heart sound signals

“…CNNs can be trained to extract more distinguishable features for the most relevant features to generate the most accurate results. To classify data, CNN architectures in 1D, 2D, 22 and 3D 23 variants can be utilized. Each method requires a large amount of data to train CNN 24 .…”