An Intrusion Detection System (IDS) is one of the fundamental building blocks in securing a network. A huge number of techniques have been proposed and implemented to improve the performance and accuracy of intrusion detection models. In recent years, efforts have been made to keep the attack surface as small as possible. However, the attack vectors have evolved in terms of complexity and diversity, and various intelligent techniques have been employed by the adversaries to exploit the ecosystems. This exploitation can either make the whole system dysfunctional or may lead to important information leakage . More specifically, with the evolution of Internet of Things (IoT), more heterogeneous and resource-constrained devices are interconnecting with each other. These devices have limited processing power and resources, particularly for detecting intrusions . Consequently, Swarm Intelligence (SI) based techniques have received considerable attention, especially in the past decade, as the SI approaches have achieved a decent success rate by optimizing various aspects of an IDS. This paper presents a systematic review with a thorough coverage of articles published between 2010 and 2020 of the state-of-the-art swarm intelligence approaches deployed in various attack surfaces for intrusion detection in various domains. The paper also presents a categorization in accordance with applicability of these SI approaches in improving various aspects of an intrusion detection process. Moreover, the paper also discusses the capabilities and features of various datasets used in experimentation. This aims to help researchers in assessing the capabilities and limitations of SI algorithms to identify security threats and challenges to design and implement an IDS for the detection of cyber attacks in various domains. Moreover, this will also help security individuals in differentiating SI based IDS with traditional ones. As such, the survey would be equally beneficial for the researchers working in the domain of swarm intelligence as well as cybersecurity. The survey highlights certain existing challenges and provides directions to address them effectively. In addition, new research directions are also identified. • A systematic review of the state-of-the-art swarm intelligence approaches deployed for intrusion detection. • A systematic effort towards classification of intrusion, weight/parameter optimization of ML-based classifiers, feature selection, and multi-objective models. • Features of various data-sets used in experimentation of swarm intelligence based IDS. • Key observations and open challenges in swarm intelligence based IDS. 

Swarm Intelligence inspired Intrusion Detection Systems - A systematic literature review

A study on initialization of connection weights of neural networks is expected to be needed because various deep neural networks based on deep learning have attracted much attention recently. However, studies on the relation between the output value of the active function and the learning performance of the neural network with respect to the connection weight value have been conducted mainly on the supervised learning model. This paper focused on improving the efficiency of autonomous neural network model by studying the connection weight initialization as the neural network model of supervised learning. Adaptive resonance theory (ART) is a major model of autonomous neural network that tries to solve the stability–plasticity dilemma by using bottom-up weights and top-down weights. The conventional weights initialization method of ART was to uniformly set all weights, but the proposed method is to initialize by using pre-trained weights. Experiments show that the ART, which initializes the connectivity weights through the proposed method, performs clustering more reliably.

The weights initialization methodology of unsupervised neural networks to improve clustering stability

Given the recognized advantages of additive manufacturing (AM) printing systems in comparison with conventional subtractive manufacturing systems, AM technology has become increasingly adopted in 3D manufacturing, with usage rates increasing dramatically. This strong growth has had a significant and direct impact not only on energy consumption but also on manufacturing time, which in turn has generated significant costs. As a result, this problem has attracted the attention of industry actors and the research community, and several studies have focused on predicting and reducing energy consumption and additive manufacturing time, which has become one of the main objectives of research in this field. However, there is no effective model yet for predicting and optimizing energy consumption and printing time in a fused deposition modeling (FDM) process while taking into account the correct part orientation that minimizes both of these costs. In this paper, a neural-network-based model has been proposed to solve this problem using experimental data from isovolumetrically shaped mechanical parts. The data will serve as the basis for proposing the appropriate model using a specific methodology based on five performance criteria with the following statistical values: R2-squared > 99%, explained variance > 99%, MAE < 0.99%, MSE < 0.02% and RMSE < 1.36%. These values show just how effective the proposed model will be in estimating energy consumption and FDM printing time, taking into account the best choice of part orientation for the lowest cost. This model provides a global understanding of the primary energy and time requirements for manufacturing while also improving the system’s cost efficiency. The results of this work can be extended and applied to other additive manufacturing processes in future work.

Modeling of Energy Consumption and Print Time for FDM 3D Printing Using Multilayer Perceptron Network

The market of modern industry is oriented towards the technology of additive manufacturing, which encompasses several processes, among which we find FDM (fused deposition modeling), which has become increasingly used due to its advantageous contribution compared to traditional techniques. Its advantages concern in particular the reduction of manufacturing time, the resistance to temperature and the limitation of human intervention, which also decreases the risks that the user of this technology may have. However, despite these advantages, FDM has some limitations, especially with regard to defects that influence the manufacturing and, of course, the quality of the final product. As a result, and because of the recognised effectiveness of machine learning algorithms in the field of FDM, we have prepared this summary paper to review recent research involving the application of these algorithms to the FDM manufacturing process. The purpose of this work is to show the applicability of ML in various tasks related to this type of process. These tasks include surface roughness prediction, part deviation detection, cost estimation and defect prediction. The results of this paper show that ML algorithms contribute effectively to achieving good prediction and classification accuracies for several tasks associated with FDM manufacturing. This work opens the door for further research to apply these ML algorithms in other tasks related to this type of process. 

Application of Machine Learning in Fused Deposition Modeling: A Review

Development and comparison of machine-learning algorithms for anomaly detection in 3D printing using vibration data

Development of Machine Learning Algorithm for Characterization and Estimation of Energy Consumption of Various Stages during 3D Printing

With the advancement of computers and its computational enhancement over several decades of use, but with the growth in the dependencies and use of these systems, more and more concerns over the risk and security issues in networks have raised. In this paper, we have proposed approach using particle swarm optimization to optimize ART. Adaptive resonance theory is one of the most well-known machine-learning-based unsupervised neural networks, which can efficiently handle high-dimensional dataset. PSO on the other hand is a swarm intelligence-based algorithm, efficient in nonlinear optimization problem and easy to implement. The method is based on anomaly detection as it can also detect unknown attack types. PSO is used to optimize vigilance parameter of ART-1 and to classify network data into attack or normal. KDD ’99 (knowledge discovery and data mining) dataset has been used for this purpose.

Optimization of Adaptive Resonance Theory Neural Network Using Particle Swarm Optimization Technique

Simulation approaches based on Deep Learning (DL) techniques have several computational stages that present information at various levels of complexity. DL has exploded in popularity over the last decade, mainly in medical image analysis, information retrieval, and computational biology. As a result, DL has effectively transformed and enhanced detection, forecast, and treatment in various medical sectors, including pathological, brain tumors, lung disease, the stomach, heart, and eye. This study introduces a unique classification fuzzy logic-based and deep neuro (DNFLT) method compared with an SVM classifier for accurate segmentation of various brain tumors. According to the experimental observations, the suggested brain tumor classification approach is more resilient than other conventional methodologies of performance indicators, sensitivities, specificity, and correctness.

Rishik Kumar

Papers

Development of Machine Learning Algorithm for Characterization and Estimation of Energy Consumption of Various Stages during 3D Printing

Optimization of Adaptive Resonance Theory Neural Network Using Particle Swarm Optimization Technique

Deep Neuro-Fuzzy Logic Technique for Brain Meningiomasa Prediction

Development and comparison of machine-learning algorithms for anomaly detection in 3D printing using vibration data