One of the major diseases that has a high mortality rate in women is breast cancer. As the death rate of women has been increasing every year, it is necessary to decrease this number to detect the cancerous cells accurately by employing various methods. This paper presents a review of various works on the detection of breast cancer using various machine learning techniques such as decision tree, random forest, K-nearest neighbour, support vector machine, logistic regression and Naïve Bayes classifier. In addition, the paper also covers various deep neural network techniques and the comparison of various works. It follows various steps, namely pre-processing of breast image, mass detection, feature selection and image segmentation, feature extraction and classification. These steps are applied on various datasets namely, Wisconsin dataset, ImageNet, BreakHis, histopathological images and MIAS. The performance of various models has been examined and made a comparative study by considering accuracy, sensitivity and specificity metrics. Authors of this paper presented an overview of the current developments in cancer research by leveraging machine learning, deep learning and transformer models. Further, the authors also proposed the future scope of the work.

We propose a new dissimilarity measure, sorting different time series and measuring their absolute and relative degree of disorganisation. This work compares this strategy with the state-of-the-art of dissimilarities or similarities measures, such as DTW, maximal information coefficient (MIC) and complexity-invariant distance (CID). Two clustering algorithms, one deterministic and one non-deterministic, K-means and hierarchical, allow us to analyse their results. To infer the accuracy, we use two different indexes, maximal HITS, and adjusted Rand index. The results of the experiments, over 128 different datasets, demonstrate that the proposed approach provides more accurate results for different domains using the proposed metrics.

Inderscience is a global company, a dynamic leading independent journal publisher disseminates the latest research across the broad fields of science, engineering and technology; management, public and business administration; environment, ecological economics and sustainable development; computing, ICT and internet/web services, and related areas.

Speech emotion recognition (SER) is a challenging and active field of collaborative, social robotics to improve human-robot interaction (HRI) and affective computing as a feedback mechanism. More recently self-supervised learning (SSL) approaches have become an important method for learning speech representations. We present results of experiments on the challenging large-scale speech emotion RAVDESS dataset. Six very large state-of-the-art self-supervised learning transformer models were trained on the speech emotion dataset. Wav2Vec2.0-XLSR-53 was the most successful of the six level-0 models and achieved classification accuracy of 93%. We propose majority voting ensemble models that combined three and five level-0 models. The five-model and three-model majority voting ensemble models achieved 96.88% and 96.53% accuracy respectively and thereby significantly outperformed the best level-0 model and surpassed the state-of-the-art.

With the progress of information technology, digital data larceny and duplicity have become very easy. Image watermarking in cryptography is a major domain that provides manifold security features like confidentiality, authenticity, integrity, etc. This research introduces a robust watermarking scheme for colour images. The proposed technique segments the colour image into three layers red, green and blue. The lifting wavelet transform (LWT) and differential histogram shifting are used to embed text watermark information into the R layer. The performance of the proposed technique was assessed using the SIPI image dataset. Test outputs show that the proposed scheme maintains the balance between imperceptibility and robustness. This scheme has a better resistance against all types of attacks like different noises, filter effects, image compressions, etc. Besides, the text watermark can be successfully extracted for different types of tampering like content removal attacks, and content addition attacks.

Frequent itemset mining has received a lot of attention in the field of data mining. Its main objective is to find groups of items that consistently appear together in datasets. Even while frequent itemset mining is useful, the algorithms for mining frequent itemsets have quite high resource requirements. In order to optimise the time and memory needs, a few improvements have been made in recent years. This study proposes CellFPM, a straightforward yet effective cellular learning automata-based method for finding frequent itemset occurrences. It works efficiently with large datasets. The efficiency of the proposed approach in time and memory requirements has been evaluated using benchmark datasets explicitly designed for performance measure. The varying size and density of the test datasets have confirmed the scalability of the suggested method. The findings show that CellFPM consistently surpasses the leading algorithms in terms of runtime and memory usage, particularly memory usage mostly.