Autonomous Underwater Vehicle (AUV) is an automated navigation device, which can independently complete the underwater investigation and detection without supervision. However, the marine environment is getting increasingly severe, and the garbage in the ocean and fishnets cast by fishermen have a great negative influence on the work of AUV. To solve the problems of low precision and slow speed in the state-of-the-art network detection of weak and small targets in underwater sonar images, Yolo5 was improved by building the multi-branch shuttle neural network The dataset is collected by multi-beam forward-looking sonar Gemini720i, and includes “fishnet” and two representative types of knitted and plastic garbage, i.e., “cloth” and “plasticbag”. The original dataset is enhanced and balanced, and the effect of dataset distribution on the model performance is studied. Utilizing pre-training, the effects of Yolo5 family, i.e., Yolo5s, Yolo5m, Yolo5l and Yolo5x are compared using the balanced dataset, to explore the impacts on detecting weak and small targets of the deeper and wider networks of this family. In addition, the combinations of Yolo5s with lightweight networks, i.e., MobileNet3 and ShuffleNet2 are experimented, which further illustrates the effectiveness of the proposed network, and can satisfy the high accuracy and real-time requirements of AUV.

Detection of Weak and Small Targets in Forward-Looking Sonar Image Using Multi-Branch Shuttle Neural Network

A kind of cyber-attack that severely paralyzes the victim server by injecting illegitimate packets of data is a DDoS attack, which is progressive in nature. Therefore its detection is a highly tedious task. Hence, IDS models are developed to detect this attack efficiently, based on machine learning algorithms such as C4.5, SVM, and KNN classifier algorithms and 10-fold cross validation techniques. The NSL-KDD bench mark dataset is employed to validate the models experimentally. A 10-fold cross validation technique is used to select the trend features, and ten trial runs are made to avoid biased output. The classic SVM classifier model reported better accuracy, but the precision and sensitivity of the C4.5 classifier algorithm are better than that of SVM and KNN models. In order to improve the performance of the machine learning based intrusion detection models, an attempt is made to feed the SVM and KNN based IDS model with the features selected by C4.5 classifier algorithm, and the obtained performance metric values are reported. It is evident from the results obtained that the hybrid combination of C4.5 with SVM out performed all other models discussed in this research with an accuracy of 0.9604.

DDoS Attack Detection Using Hybrid Machine Learning Based IDS Models

Brain tumour is one of the most complicated diseases to treat in modern medicine. In the early stages of tumour development, the radiologist’s primary concern is often an accurate and efficient study. Deep Learning has become a great tool for doctors and scientists to act decisively and on time with tumor patients. A training model that has accomplish considerable result in image detection and classification is the Deep Residual Network (ResNet) utilizing CNNs. The advancement of deep learning will assist radiologists in tumor diagnostics without the use of harmful procedures. With better understanding of MRI images, as well as increase in training speeds and accuracy, deep learning can open new doors for the medical research community. In this model, an accuracy of 95.3% is achieved across various classes of brain tumor datasets. We study the outcomes of multi class classification of brain tumour using Transfer Learning utilising pre-trained ResNet50 model using CNN architecture in this paper. 

ResNet-50 based deep neural network using transfer learning for brain tumor classification

Deep neural network (DNN) was applied in sonar image target recognition tasks, but it is very difficult to obtain enough sonar images that contain a target; as a result, the direct use of a small amount of data to train a DNN will cause overfitting and other problems. Transfer learning is the most effective way to address such scenarios. However, there is a large domain gap between optical images and sonar images, and common transfer learning methods may not be able to effectively handle it. In this paper, we propose a transfer learning method for sonar image classification and object detection called the texture feature removal network. We regard the texture features of an image as domain-specific features, and we narrow the domain gap by discarding the domain-specific features, and hence, make it easier to complete knowledge transfer. Our method can be easily embedded into other transfer learning methods, which makes it easier to apply to different application scenarios. Experimental results show that our method is effective in side-scan sonar image classification tasks and forward-looking sonar image detection tasks. For side-scan sonar image classification tasks, the classification accuracy of our method is enhanced by 4.5% in a supervised learning experiment, and for forward-looking sonar detection tasks, the average precision (AP) is also significantly improved.

https://www.mdpi.com/2072-4292/15/3/616/pdf?version=1674204642

A Texture Feature Removal Network for Sonar Image Classification and Detection

Existing SYN flood attack detection methods have obvious problems such as poor feature selectivity, weak generalization ability, easy overfitting, and low accuracy during training. In the paper, we present a SYN flood attack detection method based on the Hierarchical Multihad Self-Attention (HMHSA) mechanism. First, we use one-hot encoding and normalization to preprocess traffic data. Then the preprocessed traffic data is transmitted to the Feature-based Multihead Self-Attention (FBMHA) layer for feature selection. Finally, we use data slices to determine the features of the preprocessed traffic data under time series by passing the preprocessed traffic data into the Slice-based Multihead Self-Attention (SBMHA) layer. We tested the proposed method on different datasets. The experimental results show that compared with other works, our method presents better in feature selection and higher detection accuracy (even up to 99.97%).

/pdf/a-syn-flood-attack-detection-method-based-on-hierarchical-3cemwuqx.pdf

A SYN Flood Attack Detection Method Based on Hierarchical Multihead Self-Attention Mechanism

This paper focus on the development of a Brain Computer Interface (BCI) using neural network, where BCI is an electronic communication system which accomplish the operating principle of “stimulate thinking and make it happen destitute of anyone's physical efforts”. Based on the environmental conditions the BCI allows the users to act on his thoughts, without using their peripheral muscles and nerves. A machine learning algorithm is included in all the neighboring BCI, which learns a function from the training data's besides used to distinguish non identical brain activities. The investigation of this work with Brain Computer Interface (BCI) signals possessed from Electro-EncephaloGraphy (EEG) to bring out a relationship between a person's mind condition and a computer employed signal processing system that illustrates the EEG signals. A Radial basic function framework for machine learning is used here for the linear discriminant analysis of EEG data. Based on the Radial Basic Function's (RBF) the output an intended mental task can be performed.

Development of Brain Computer Interface, using Neural Network

Side scan sonar image augmentation for sediment classification using deep learning based transfer learning approach

Deep learning binary fruit fly algorithm for identifying SYN flood attack from TCP/IP

In present world, it is highly observed that there is a steep decline in the agricultural sector of India. This decline is mainly due to loss of soil fertility and Depletion of ground Water table as a result of excessive use of fertilizers & pesticides, drought conditions arised as a result of failure of monsoon and unavailability of agricultural lands due to increased population and demand for food. To overcome these conditions a new form of cultivation technique called hydroponics can be used. The way toward developing plants in water without soil is alluded to as aquaculture. Required plant sustenance is given by water and supplement arrangements. The principle maxim of this paper is to develop various harvests whose ideal qualities are put away in cloud, if any of the boundaries goes sudden or low it is constrained by node mcu which is interfaced with IOT. The need of IOT is to send the recovered information to the web (mass stockpiling) and PC is accustomed to imparting the current status to the client through web with the goal that checking and support will get simpler. Asset the board in a tank-farming hardware would turn out to be not so much convoluted but rather more productive dependent on the investigation results.

D. Ravikumar

Papers

Development of Brain Computer Interface, using Neural Network

Side scan sonar image augmentation for sediment classification using deep learning based transfer learning approach

Deep learning binary fruit fly algorithm for identifying SYN flood attack from TCP/IP

IoT based Automation of Hydroponics using Node MCU Interface

Analysis of silica based prosthetic human teeth using 3D printing machine for dental assistance