Cephalopods Classification Using Fine Tuned Lightweight Transfer Learning Models

Cephalopods identification is a formidable task that involves hand inspection and close observation by a malacologist.Manual observation and iden-tification take time and are always contingent on the involvement of experts.A system is proposed to alleviate this challenge that uses transfer learning techni-ques to classify the cephalopods automatically.In the proposed method,only the Lightweight pre-trained networks are chosen to enable IoT in the task of cephalopod recognition.First,the efficiency of the chosen models is determined by evaluating their performance and comparing thefindings.Second,the models arefine-tuned by adding dense layers and tweaking hyperparameters to improve the classification of accuracy.The models also employ a well-tuned Rectified Adam optimizer to increase the accuracy rates.Third,Adam with Gradient Cen-tralisation(RAdamGC)is proposed and used infine-tuned models to reduce the training time.The framework enables an Internet of Things(IoT)or embedded device to perform the classification tasks by embedding a suitable lightweight pre-trained network.Thefine-tuned models,MobileNetV2,InceptionV3,and NASNet Mobile have achieved a classification accuracy of 89.74%,87.12%,and 89.74%,respectively.Thefindings have indicated that thefine-tuned models can classify different kinds of cephalopods.The results have also demonstrated that there is a significant reduction in the training time with RAdamGC.

Ultra-lightweight CNN design based on neural architecture search and knowledge distillation: A novel method to build the automatic recognition model of space target ISAR images

In this paper,a novel method of ultra-lightweight convolution neural network(CNN)design based on neural architecture search(NAS)and knowledge distillation(KD)is proposed.It can realize the automatic construction of the space target inverse synthetic aperture radar(ISAR)image recognition model with ultra-lightweight and high accuracy.This method introduces the NAS method into the radar image recognition for the first time,which solves the time-consuming and labor-consuming problems in the artificial design of the space target ISAR image automatic recognition model(STIIARM).On this basis,the NAS model’s knowledge is transferred to the student model with lower computational complexity by the flow of the solution procedure(FSP)distillation method.Thus,the decline of recognition accuracy caused by the direct compression of model structural parameters can be effectively avoided,and the ultralightweight STIIARM can be obtained.In the method,the Inverted Linear Bottleneck(ILB)and Inverted Residual Block(IRB)are firstly taken as each block’s basic structure in CNN.And the expansion ratio,output filter size,number of IRBs,and convolution kernel size are set as the search parameters to construct a hierarchical decomposition search space.Then,the recognition accuracy and computational complexity are taken as the objective function and constraint conditions,respectively,and the global optimization model of the CNN architecture search is established.Next,the simulated annealing(SA)algorithm is used as the search strategy to search out the lightweight and high accuracy STIIARM directly.After that,based on the three principles of similar block structure,the same corresponding channel number,and the minimum computational complexity,the more lightweight student model is designed,and the FSP matrix pairing between the NAS model and student model is completed.Finally,by minimizing the loss between the FSP matrix pairs of the NAS model and student model,the student model’s weight adjustment is completed.Thus the ultra-lightweight and high accuracy STIIARM is obtained.The proposed method’s effectiveness is verified by the simulation experiments on the ISAR image dataset of five types of space targets.

Two-Stage Edge-Side Fault Diagnosis Method Based on Double Knowledge Distillation

With the rapid development of the Internet of Things(IoT),the automation of edge-side equipment has emerged as a significant trend.The existing fault diagnosismethods have the characteristics of heavy computing and storage load,and most of them have computational redundancy,which is not suitable for deployment on edge devices with limited resources and capabilities.This paper proposes a novel two-stage edge-side fault diagnosis method based on double knowledge distillation.First,we offer a clustering-based self-knowledge distillation approach(Cluster KD),which takes the mean value of the sample diagnosis results,clusters them,and takes the clustering results as the terms of the loss function.It utilizes the correlations between faults of the same type to improve the accuracy of the teacher model,especially for fault categories with high similarity.Then,the double knowledge distillation framework uses ordinary knowledge distillation to build a lightweightmodel for edge-side deployment.We propose a two-stage edge-side fault diagnosismethod(TSM)that separates fault detection and fault diagnosis into different stages:in the first stage,a fault detection model based on a denoising auto-encoder(DAE)is adopted to achieve fast fault responses;in the second stage,a diverse convolutionmodel with variance weighting(DCMVW)is used to diagnose faults in detail,extracting features frommicro andmacro perspectives.Through comparison experiments conducted on two fault datasets,it is proven that the proposed method has high accuracy,low delays,and small computation,which is suitable for intelligent edge-side fault diagnosis.In addition,experiments show that our approach has a smooth training process and good balance.

A Deep Learning-Based Crowd Counting Method and System Implementation on Neural Processing Unit Platform

In this paper, a deep learning-based method is proposed for crowdcountingproblems. Specifically, by utilizing the convolution kernel densitymap, the ground truth is generated dynamically to enhance the featureextractingability of the generator model. Meanwhile, the “cross stage partial”module is integrated into congested scene recognition network (CSRNet) toobtain a lightweight network model. In addition, to compensate for the accuracydrop owing to the lightweight model, we take advantage of “structuredknowledge transfer” to train the model in an end-to-end manner. It aimsto accelerate the fitting speed and enhance the learning ability of the studentmodel. The crowd-counting system solution for edge computing is alsoproposed and implemented on an embedded device equipped with a neuralprocessing unit. Simulations demonstrate the performance improvement ofthe proposed solution in terms of model size, processing speed and accuracy.The performance on the Venice dataset shows that the mean absolute error(MAE) and the root mean squared error (RMSE) of our model drop by32.63% and 39.18% compared with CSRNet. Meanwhile, the performance onthe ShanghaiTech PartB dataset reveals that the MAE and the RMSE of ourmodel are close to those of CSRNet. Therefore, we provide a novel embeddedplatform system scheme for public safety pre-warning applications.

Identification of Weather Phenomena Based on Lightweight Convolutional Neural Networks

Weather phenomenon recognition plays an important role in the field of meteorology.Nowadays,weather radars and weathers sensor have been widely used for weather recognition.However,given the high cost in deploying and maintaining the devices,it is difficult to apply them to intensive weather phenomenon recognition.Moreover,advanced machine learning models such as Convolutional Neural Networks(CNNs)have shown a lot of promise in meteorology,but these models also require intensive computation and large memory,which make it difficult to use them in reality.In practice,lightweight models are often used to solve such problems.However,lightweight models often result in significant performance losses.To this end,after taking a deep dive into a large number of lightweight models and summarizing their shortcomings,we propose a novel lightweight CNNs model which is constructed based on new building blocks.The experimental results show that the model proposed in this paper has comparable performance with the mainstream non-lightweight model while also saving 25 times of memory consumption.Such memory reduction is even better than that of existing lightweight models.

MobileNet-Based IoT Malware Detection with Opcode Features

In recent years,with the rapid development of Internet and hardware technologies,the number of Internet of things(IoT)devices has grown exponentially.However,IoT devices are constrained by power consumption,making the security of IoT vulnerable.Malware such as Botnets and Worms poses significant security threats to users and enterprises alike.Deep learning models have demonstrated strong performance in various tasks across different domains,leading to their application in malicious software detection.Nevertheless,due to the power constraints of IoT devices,the well-performanced large models are not suitable for IoT malware detection.In this paper we propose a malware detection method based on Markov images and MobileNet,offering a cost-effective,efficient,and high-performing solution for malware detection.Additionally,this paper innovatively analyzes the robustness of opcode sequences.

Target localization method of non-cooperative spacecraft on on-orbit service

With the explosion of the number of meteoroid/orbital debris in terrestrial space in recent years, the detection environment of spacecraft becomes more complex. This phenomenon causes most current detection methods based on machine learning intractable to break through the two difficulties of solving scale transformation problem of the targets in image and accelerating detection rate of high-resolution images. To overcome the two challenges, we propose a novel noncooperative target detection method using the framework of deep convolutional neural network.Firstly, a specific spacecraft simulation dataset using over one thousand images to train and test our detection model is built. The deep separable convolution structure is applied and combined with the residual network module to improve the network’s backbone. To count the different shapes of the spacecrafts in the dataset, a particular prior-box generation method based on K-means cluster algorithm is designed for each detection head with different scales. Finally, a comprehensive loss function is presented considering category confidence, box parameters, as well as box confidence. The experimental results verify that the proposed method has strong robustness against varying degrees of luminance change, and can suppress the interference caused by Gaussian noise and background complexity. The mean accuracy precision of our proposed method reaches 93.28%, and the global loss value is 13.252. The comparative experiment results show that under the same epoch and batchsize, the speed of our method is compressed by about 20% in comparison of YOLOv3, the detection accuracy is increased by about 12%, and the size of the model is reduced by nearly 50%.

PokerNet:Expanding Features Cheaply via Depthwise Convolutions

Pointwise convolution is usually utilized to expand or squeeze features in modern lightweight deep models.However,it takes up most of the overall computational cost(usually more than 90%).This paper proposes a novel Poker module to expand features by taking advantage of cheap depthwise convolution.As a result,the Poker module can greatly reduce the computational cost,and meanwhile generate a large number of effective features to guarantee the performance.The proposed module is standardized and can be employed wherever the feature expansion is needed.By varying the stride and the number of channels,different kinds of bottlenecks are designed to plug the proposed Poker module into the network.Thus,a lightweight model can be easily assembled.Experiments conducted on benchmarks reveal the effectiveness of our proposed Poker module.And our Poker Net models can reduce the computational cost by 7.1%-15.6%.Poker Net models achieve comparable or even higher recognition accuracy than previous state-of-the-art(SOTA)models on the Image Net ILSVRC2012 classification dataset.Code is available at https://github.com/diaomin/pokernet.