The advancement in the machine learning and the computer vision has caused several improvements and development in numerous of domains. Capsule neural networks are one such machine learning system that imitates the neural system and develops the structures based on the hierarchical relationships. It does the inverse operation of the computer graphic in representing an object by, segregating the object in the image into different part and viewing the in-existing relationship between the each parts to represent in order to preserve even the minute details related to the object, unlike CNN that losses major of the information’s related to the spatial location of the object that are essential in the segmentation and the detection. So the paper presents the comparative study of the capsule neural network in various application, presenting the efficiency of the capsules networks over the convolutional neural networks.

https://www.irojournals.com/aicn/V1/I1/03.pdf

Comparative study of capsule neural network in various applications

Governments around the globe are promoting smart city applications to enhance the quality of daily-life activities in urban areas. Smart cities include internet-enabled devices that are used by applications like health care, power grid, water treatment, traffic control, etc to enhance its effectiveness. The expansion in the quantity of Internet-of-things (IoT) based botnet attacks is due to the growing trend of Internet-enabled devices. To provide advanced cyber security solutions to IoT devices and smart city applications, this paper proposes a deep learning (DL) based botnet detection system that works on network traffic flows. The botnet detection framework collects the network traffic flows, converts them into connection records and uses a DL model to detect attacks emanating from the compromised IoT devices. To determine an optimal DL model, many experiments are conducted on well-known and recently released benchmark data sets. Further, the datasets are visualized to understand its characteristics. The proposed DL model outperformed the conventional machine learning (ML) models.

Network Flow based IoT Botnet Attack Detection using Deep Learning

Multiclass malware classification via first- and second-order texture statistics

Waste generation has been increasing drastically based on the world’s population and economic growth. This has significantly affected human health, natural life, and ecology. The utilization of limited natural resources, and the harming of the earth in the process of mineral extraction, and waste management have far exceeded limits. The recycling rate are continuously increasing; however, assessments show that humans will be creating more waste than ever before. Some difficulties during recycling include the significant expense involved during the separation of recyclable waste from non-disposable waste. Machine learning is the utilization of artificial intelligence (AI) that provides a framework to take as a structural improvement of the fact without being programmed. Machine learning concentrates on the advancement of programs that can obtain the information and use it to learn to make future decisions. The classification and separation of materials in a mixed recycling application in machine learning is a division of AI that is playing an important role for better separation of complex waste. The primary purpose of this study is to analyze AI by focusing on machine learning algorithms used in recycling systems. This study is a compilation of the most recent developments in machine learning used in recycling industries.

Applying machine learning approach in recycling

With the tremendous growth of the internet, cyberspace is facing several threats from the attackers. Threats like spam emails account for 55% of total emails according to the Symantec monthly threat report. Over time, the attackers moved on to image spam to evade the text-based spam filters. To deal with this, the researchers have several machine learning and deep learning approaches that use various features like metadata, color, shape, texture features. But the Deep Convolutional Neural Network (DCNN) and transfer learning-based pre-trained CNN models are not explored much for Image spam classification. Therefore, in this work, 2 DCNN models along with few pre-trained ImageNet architectures like VGG19, Xception are trained on 3 different datasets. The effect of employing a Cost-sensitive learning approach to handle data imbalance is also studied. Some of the proposed models in this work achieves an accuracy up to 99% with zero false positive rate in best case.

Deep Convolutional Neural Network Based Image Spam Classification

Building an image processing model for prediction or classification application has to overcome quite a lot of challenges. Convolutional neural network (CNN) is the pillar of image processing in deep learning perspective. In order to bring down the disadvantages and for improving the performance compared to the CNN, a new architecture of CNN had been devised which is known as Capsule neural network (Capsule-Net). By this paper we analyze Capsule-Net for solid waste management which is separation of plastic and non-plastic. This task is viewed as of at most significance in today’s world due to volumes of waste generated and nonavailability of human labor for this work. The capsule-Net is evaluated using 2 different datasets. Dataset 1 represents materials collected from public places and Dataset 2 represents materials collected from private environment. The proposed architecture with capsule-Net gives an accuracy of 96.3% for Dataset 1 and 95.7% for Dataset 2. The necessary hardware setup has been developed and tested. This will be a grace to the society which faces unexplainable difficulty in disposing wastes. It is inexpensive labor free and harmless to health.

Capsule Neural Networks and Visualization for Segregation of Plastic and Non-Plastic Wastes

Real-time prediction of domain names that are generated using the Domain Generation Algorithms (DGAs) is a challenging cyber security task. Scope to collect the vast amount of data for training favored data-driven techniques and deep learning architectures have the potential to address this challenge. This paper proposes a deep learning framework using long short-term memory (LSTM) architecture for prediction of the domain names that are generated using the DGAs. Binary classification had benign and DGA domain names and multiclass classification was performed using 20 different DGAs. For the binary classification, LSTM model gave accuracy of 98.7% and 71.3% on two different test data sets and for the multi-class classification, it gave accuracy of 68.3% and 67.0% respectively. Two diversified data sets were used to analyze the robustness of the LSTM architecture.

Deep Learning Framework for Domain Generation Algorithms Prediction Using Long Short-term Memory

Spamming and Phishing attacks are the most common security challenges we face in today’s cyber world. The existing methods for the Spam and Phishing detection are based on blacklisting and heuristics technique. These methods require human intervention to update if any new Spam and Phishing activity occurs. Moreover, these are completely inefficient in detecting new Spam and Phishing activities. These techniques can detect malicious activity only after the attack has occurred. Machine learning has the capability to detect new Spam and Phishing activities. This requires extensive domain knowledge for feature learning and feature representation. Deep learning is a method of machine learning which has the capability to extract optimal feature representation from various samples of benign, Spam and Phishing activities by itself. To leverage, this work uses various deep learning architectures for both Spam and Phishing detection with electronic mail (Email) and uniform resource locator (URL) data sources. Because in recent years both Email and URL resources are the most commonly used by the attackers to spread malware. Various datasets are used for conducting experiments with deep learning architectures. For comparative study, classical machine learning algorithms are used. These datasets are collected using public and private data sources. All experiments are run till 1,000 epochs with varied learning rate 0.01–0.5. For comparative study various classical machine learning classifiers are used with domain level feature extraction. For deep learning architectures and classical machine learning algorithms to convert text data into numeric representation various natural language processing text representation methods are used. As far as anyone is concerned, this is the first attempt, a framework that can examine and connect the occasions of Spam and Phishing activities from Email and URL sources at scale to give cyber threat situational awareness. The created framework is exceptionally versatile and fit for distinguishing the malicious activities in close constant. In addition, the framework can be effectively reached out to deal with vast volume of other cyber security events by including extra resources. These qualities have made the proposed framework emerge from some other arrangement of comparative kind.

Deep Learning Framework for Cyber Threat Situational Awareness Based on Email and URL Data Analysis

In this paper we propose a deep learning framework for classification of malware. There has been an enormous increase in the volume of malware generated lately which represents a genuine security danger to organizations and people. So as to battle the expansion of malwares, new strategies are needed to quickly identify and classify malware. Malimg dataset, a publicly available benchmark data set was used for the experimentation. The architecture used in this work is a hybrid cost-sensitive network of one-dimensional Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) network which obtained an accuracy of 94.4%, an increase in performance compared to work done by [1] which got 84.9%. Hyper parameter tuning is done on deep learning architecture to set the parameters. A learning rate of 0.01 was taken for all experiments. Train-test split of 70-30% was done during experimentation. This facilitates to find how well the models perform on imbalanced data sets. Usual methods like disassembly, decompiling, de-obfuscation or execution of the binary need not be done in this proposed method. The source code and the trained models are made publicly available for further research.

Deep Learning Framework and Visualization for Malware Classification

Machine learning has played an important role in the last decade mainly in natural language processing, image processing and speech recognition where it has performed well in comparison to the classical rule based approach. The machine learning approach has been used in cyber security use cases namely, intrusion detection, malware analysis, traffic analysis, spam and phishing detection etc. Recently, the advancement of machine learning typically called as ‘deep learning’ outperformed humans in several long standing artificial intelligence tasks. Deep learning has the capability to learn optimal feature representation by itself and more robust in an adversarial environment in compared to classical machine learning algorithms. This approach is in early stage in cyber security. In this work, to leverage the application of deep learning architectures towards cyber security, we consider intrusion detection, traffic analysis and Android malware detection. In all the experiments of intrusion detection, deep learning architectures performed well in compared to classical machine learning algorithms. Moreover, deep learning architectures have achieved good performance in traffic analysis and Android malware detection too.

S. Akarsh

Papers

Capsule Neural Networks and Visualization for Segregation of Plastic and Non-Plastic Wastes

Deep Learning Framework for Domain Generation Algorithms Prediction Using Long Short-term Memory

Deep Learning Framework for Cyber Threat Situational Awareness Based on Email and URL Data Analysis

Deep Learning Framework and Visualization for Malware Classification

Application of Deep Learning Architectures for Cyber Security