As convolutional neural network contains many redundant parameters, a lot of methods have been developed to compress the network for accelerating inference. Among these, network pruning, which is a kind of widely used...
详细信息
ISBN:
(数字)9781728177090
ISBN:
(纸本)9781728177106
As convolutional neural network contains many redundant parameters, a lot of methods have been developed to compress the network for accelerating inference. Among these, network pruning, which is a kind of widely used approaches, can effectively decrease the memory capacity and reduce the computation cost. Herein, we propose a competitive pruning approach based on Soft Filter Pruning (SFP) by taking account of the scaling factors y of Batch Normalization (BN) layers as the criterion of filter selection strategy. During the soft pruning procedure, in each epoch only y values of BN layers less than threshold are set to zero instead of setting the weights of selected filters in convolutional layers to zero. Compared to the existing approaches, the proposed method can obtain a highly increased accuracy on image recognition. Notably, on CIFAR-10, the proposed method reduces the same 40.8% FLOPs as SFP on ResNet-110 with even 0.87% top-1 accuracy improvement.
Detection of the objects around a vehicle is important for a safe and successful navigation of an autonomous vehicle. Instance segmentation provides a fine and accurate classification of the objects such as cars, truc...
详细信息
ISBN:
(数字)9781728177090
ISBN:
(纸本)9781728177106
Detection of the objects around a vehicle is important for a safe and successful navigation of an autonomous vehicle. Instance segmentation provides a fine and accurate classification of the objects such as cars, trucks, pedestrians, etc. In this study, we propose a fast and accurate approach which can detect and segment the object instances which can be adapted to new conditions without requiring the labels from the new condition. Furthermore, the performance of the instance segmentation does not degrade in detection of the objects in the original condition after it adapts to the new condition. To our knowledge, currently there are not other methods which perform unsupervised domain adaptation for the task of instance segmentation using non-synthetic datasets. We evaluate the adaptation capability of our method on two datasets. Firstly, we test its capacity of adapting to a new domain; secondly, we test its ability to adapt to new weather conditions. the results show that it can adapt to new conditions with an improved accuracy while preserving the accuracy of the original condition.
Remote scene classification serves a vital role in many applications. However, satellite images are often blurred and degraded due to aerosol scattering under fog, haze, and other weather conditions, reducing the imag...
详细信息
ISBN:
(数字)9781728177090
ISBN:
(纸本)9781728177106
Remote scene classification serves a vital role in many applications. However, satellite images are often blurred and degraded due to aerosol scattering under fog, haze, and other weather conditions, reducing the image contrast and color fidelity. State-of-the-art remote sensing classification models building upon convolutional neural networks (CNNs) are mostly trained on annotated datasets of clear satellite images. When applied to blurred images, they will suffer a great degradation in performance. To address this problem, we adopt the domain adaptation algorithm TADA and propose Transferable Attention enhanced Adversarial Adaptation Network (TA 3 N), which utilizes annotated data in clear images by applying knowledge transferring from clear image domain to blurred image domain. Our TA 3 N first integrates spatial attention to focus on salient areas which are discriminative and transferable. In addition, domain discriminator and adversarial training via gradient reversal layer are used to minimize the discrepancies in extracted features from clear and degraded domains. We synthesize degraded remote scene classification dataset SSI based on FoHIS model. Experiments on degraded SSI showed that TA 3 N significantly outperforms baseline and other state-of-the-art domain adaptation methods.
Deep Learning (DL), a groundbreaking branch of Machine Learning (ML), has emerged as a driving force in boththeoretical and applied Artificial Intelligence (AI). DL algorithms, rooted in complex and non-linear artifi...
详细信息
Deep Learning (DL), a groundbreaking branch of Machine Learning (ML), has emerged as a driving force in boththeoretical and applied Artificial Intelligence (AI). DL algorithms, rooted in complex and non-linear artificial neural systems, excel at extracting high-level features from data. DL has demonstrated human -level performance in real-world tasks, including clinical diagnostics, and has unlocked solutions to previously intractable problems in virtual agent design, robotics, genomics, neuroimaging, computer vision, and industrial automation. In this paper, the most relevant advances from the last few years in Artificial Intelligence (AI) and several applications to neuroscience, neuroimaging, computer vision, and robotics are presented, reviewed and discussed. In this way, we summarize the state-of-the-art in AI methods, models and applications within a collection of works presented at the 9thinternationalconference on the Interplay between Natural and Artificial Computation (IWINAC). the works presented in this paper are excellent examples of new scientific discoveries made in laboratories that have successfully transitioned to real-life applications.
this paper mainly introduces the modeling based on Cosserat rod theory and focuses on the adaptive neural network controller design based on model. In dealing withthe external interference withthe environment and th...
详细信息
ISBN:
(数字)9781728162461
ISBN:
(纸本)9781665419352
this paper mainly introduces the modeling based on Cosserat rod theory and focuses on the adaptive neural network controller design based on model. In dealing withthe external interference withthe environment and the unmodeled dynamics of the system, a neural network (NN) is introduced to compensate it, and using Backstepping method to design the adaptive controller, finally the stability of the closed-loop system and the convergence of the signal in the system is proved with Lyapunov function theory, ensuring the end of the manipulator can track the given signal. In addition, the simulation experiment of soft manipulator swing and constant angle tracking control are carried out, and the simulation shows the rationality of the proposed controller.
State-of-charge (SOC) estimation is the key to the BMS of lithium-ion batteries. At present, data-driven methods are more popularly used to estimate the SOC. Among them, the deep learning technology especially recurre...
详细信息
ISBN:
(数字)9781728177090
ISBN:
(纸本)9781728177106
State-of-charge (SOC) estimation is the key to the BMS of lithium-ion batteries. At present, data-driven methods are more popularly used to estimate the SOC. Among them, the deep learning technology especially recurrent neural networks (RNNs) performed very well. In this paper, we proposed a bidirectional long short-term memory (Bi-LSTM) neural network to perform SOC estimation. the bidirectional LSTM layer is employed to catch the forward and backward temporal dependencies of battery sequential states, and shows better performance in SOC estimation of LiBs. With a public dataset of vehicle driving cycles, we separately train our proposed network at three different constant temperatures (0 °C, 10 °C, and 25 °C), and the evaluated MAEs are 0.498%, 0.411 %, 0.738%. Besides, the Bi-LSTM network trained with overall datasets at three temperatures achieves a mean absolute error (MAE) of 0.616% and maximum absolute error (MaxE) of 3.809%. It shows that the proposed method is robust and accurate in SOC estimation.
Stress is an increasingly common problem. thanks to the use of various mathematical methods, it can be described mathematically based on the EEG signal. Generally, the stress in mathematical analysis can be divided in...
详细信息
Stress is an increasingly common problem. thanks to the use of various mathematical methods, it can be described mathematically based on the EEG signal. Generally, the stress in mathematical analysis can be divided into several models. To determine the variability in the stress-related EEG signal, the periodograms used for the overall assessment are checked.
A general tracking problem is formulated and solved for a third-order Duffing-Holmes type chaotic oscillator implemented as an electronic circuit with unknown parameters. Two nonlinear controllers, based on adaptive b...
详细信息
A general tracking problem is formulated and solved for a third-order Duffing-Holmes type chaotic oscillator implemented as an electronic circuit with unknown parameters. Two nonlinear controllers, based on adaptive backstepping techniques, are derived and compared. Both are sufficiently robust and accurate for practical implementation. the same approach may be used for chaos stabilization and synchronization.
In this paper the lossless block encoder application using the adaptive Golomb code and asymmetric inter-channel dependencies are presented. Using the sets of settings obtained from the study of inter-channel dependen...
详细信息
In this paper the lossless block encoder application using the adaptive Golomb code and asymmetric inter-channel dependencies are presented. Using the sets of settings obtained from the study of inter-channel dependencies, the lowest bit averages were found within the channels. In the next step the algorithm was improved by adaptation within the blocks, what allowed to improving the degree of compression.
Stabilization of a linear discrete-time large-scale interconnected systems composed of identical subsystems is studied. the controls of every subsystem are delayed. the control design is based on a state transformatio...
详细信息
Stabilization of a linear discrete-time large-scale interconnected systems composed of identical subsystems is studied. the controls of every subsystem are delayed. the control design is based on a state transformation that decouples the subsystems. then, a suitable design method is used. Robustness to deal with systems with uncertainties is guaranteed. the results are illustrated by an example.
暂无评论