Cybersecurity has received significant attention globally, with the ever-continuing expansion of internet usage, due to growing trends and adverse impacts of cybercrimes, which include disrupting businesses, corrupting or altering sensitive data, stealing or exposing information, and illegally accessing a computer network. As a popular way, different kinds of firewalls, antivirus systems, and Intrusion Detection Systems (IDS) have been introduced to protect a network from such attacks. Recently, Machine Learning (ML), including Deep Learning (DL) based autonomous systems, have been state-of-the-art in cyber security, along with their drastic growth and superior performance. This study aims to develop a novel IDS system that gives more attention to classifying attack cases correctly and categorizes attacks into subclass levels by proposing a two-step process with a cascaded framework. The proposed framework recognizes the attacks using one ML model and classifies them into subclass levels using the other ML model in successive operations. The most challenging part is to train both models with unbalanced cases of attacks and non-attacks in the datasets, which is overcome by proposing a data augmentation technique. Precisely, limited attack samples of the dataset are augmented in the training set to learn the attack cases properly. Finally, the proposed framework is implemented with NN, the most popular ML model, and evaluated with the NSL-KDD dataset by conducting a rigorous analysis of each subclass emphasizing the major attack class. The proficiency of the proposed cascaded approach with data augmentation is compared with the other three models: the cascaded model without data augmentation and the standard single NN model with and without the data augmentation technique. Experimental results on the NSL-KDD dataset have revealed the proposed method as an effective IDS system and outperformed existing state-of-the-art ML models. 

Integration of skin color property in face detection algorithm is a recent trend to improve accuracy. The existing skin color matching techniques are illumination condition dependent, which directly impacts the face detection algorithm. In this study, a novel illumination condition invariant skin color matching method is proposed which is a composite of two rules to balance the high and low intensity facial images by individual rule. The proposed skin color matching method is incorporated into Haar Feature based Face Detection (HFFD) algorithm for face detection and is verified on a large set of images having variety of skin colors and also varying illumination intensities. Experimental results reveal the effectiveness and robustness of the proposed method outperforming other existing methods.

Handwritten numeral recognition (HNR) has gained much attention in present days as it can be applied in range of applications. Research on unconstrained HNR has shown impressive progress in few scripts but is far behind for Bangla although it is one of the major languages. Bangla contains similar shaped numerals which are difficult to distinguish even in printed form and this makes Bangla HNR (BHNR) a challenging task. Our goal in this study is to build up a superior BHNR framework and consequently explore the profound design of Long Short Term Memory (LSTM) method. LSTM is a variation of Recurrent Neural Network and is effectively used for sequence ordering with its distinct features. This study considered deep architecture of LSTM for better performance. The proposed BHNR with deep LSTM (BNHR-DLSTM) standardizes the composed numeral images first and then utilizes two layers of LSTM to characterize singular numerals. Benchmark dataset with 22000 handwritten numerals having various shapes, sizes and varieties are utilized to examine the proficiency of BNHR-DLSTM. The proposed method indicates agreeable recognition precision and beat other conspicuous existing methods.

The conventional real-valued neuro-fuzzy method (RNF) is based on classic fuzzy systems with antecedent membership functions and consequent singletons. Rules in RNF are made by all the combinations of membership functions; thus, the number of rules as well as total parameters increase rapidly with the number of inputs. Although network parameters are relatively less in the recently developed complex-valued neuro-fuzzy (CVNF) and quaternion neuro-fuzzy (QNF), parameters increase with number of inputs. This study investigates simplified fuzzy rules that constrain rapid increment of rules with inputs; and proposed simplified RNF (SRNF), simplified CVNF (SCVNF) and simplified QNF (SQNF) employing the proposed simplified fuzzy rules in conventional methods. The proposed simplified neuro-fuzzy learning methods differ from the conventional methods in their fuzzy rule structures. The methods tune fuzzy rules based on the gradient descent method. The number of rules in these methods are equal to the number of divisions of input space; and hence they require significantly less number of parameters to be tuned. The proposed methods are tested on function approximations and classification problems. They exhibit much less execution time than the conventional counterparts with equivalent accuracy. Due to less number of parameters, the proposed methods can be utilized for the problems (e.g., real-time control of large systems) where the conventional methods are difficult to apply due to time constrain.

Recognition of handwritten numerals has gained much interest in recent years due to its various potential applications. Bengali is the fifth ranked among the spoken languages of the world. However, due to inherent difficulties of Bengali numeral recognition, a very few study on handwritten Bengali numeral recognition is found with respect to other major languages. The existing Bengali numeral recognition methods used distinct feature extraction techniques and various classification tools. Recently, convolutional neural network (CNN) is found efficient for image classification with its distinct features. In this paper, we have investigated a CNN based Bengali handwritten numeral recognition scheme. Since English numerals are frequently used with Bengali numerals, handwritten Bengali-English mixed numerals are also investigated in this study. The proposed scheme uses moderate pre-processing technique to generate patterns from images of handwritten numerals and then employs CNN to classify individual numerals. It does not employ any feature extraction method like other related works. The proposed method showed satisfactory recognition accuracy on the benchmark data set and outperformed other prominent existing methods for both Bengali and Bengali-English mixed cases. 

Recently, speech recognition (SR) has drawn a great attraction to the research community due to its importance in human-computer interaction bearing scopes in many important tasks. In a SR system, acoustic modelling (AM) is crucial one which contains statistical representation of every distinct sound that makes up the word. A number of prominent SR methods are available for English and Russian languages with Deep Belief Network (DBN) and other techniques with respect to other major languages such as Bangla. This paper investigates acoustic modeling of Bangla words using DBN combined with HMM for Bangla SR. In this study, Mel Frequency Cepstral Coefficients (MFCCs) is used to accurately represent the shape of the vocal tract that manifests itself in the envelope of the short time power spectrum. Then DBN is trained with these feature vectors to calculate each of the phoneme states. Later on enhanced gradient is used to slightly adjust the model parameters to make it more accurate. In addition, performance on training RBMs improved by using adaptive learning, weight decay and momentum factor. Total 840 utterances (20 utterances for each of 42 speakers) of the words are used in this study. The proposed method is shown satisfactory recognition accuracy and outperformed other prominent existing methods.

Handwritten character recognition complexity varies among different languages due to distinct shapes, strokes and number of characters. Numerous works in handwritten character recognition are available for English with respect to other major languages such as Bangla. Existing methods use distinct feature extraction techniques and various classification tools in their recognition schemes. Recently, Convolutional Neural Network (CNN) is found efficient for English handwritten character recognition. In this paper, a CNN based Bangla handwritten character recognition is investigated. The proposed method normalizes the written character images and then employ CNN to classify individual characters. It does not employ any feature extraction method like other related works. 20000 handwritten characters with different shapes and variations are used in this study. The proposed method is shown satisfactory recognition accuracy and outperformed some other prominent exiting methods.

Algorithms inspired from natural phenomena are seem to be efficient to solve various optimization problems. This paper investigates a new technique inspiring from the animal group living behavior to solve traveling salesman problem (TSP), the most popular combinatorial optimization problem. The proposed producer-scrounger method (PSM) models roles and interactions of three types of animal group members: producer, scrounger and dispersed. PSM considers a producer having the best tour, few dispersed members having worse tours and scroungers. In each iteration, the producer scans for better tour, scroungers explore new tours while moving toward producer’s tour; and dispersed members randomly checks new tours. For producer’s scanning, PSM randomly selects a city from the producer’s tour and rearranges its connection with several near cities for better tours. Swap operator and swap sequence based operation is employed in PSM to update a scrounger towards the producer. The proposed PSM has been tested on a large number of benchmark TSPs and outcomes compared to genetic algorithm and ant colony optimization. Experimental results revealed that proposed PSM is a good technique to solve TSP providing the best tours in most of the TSPs.