Название: Modern Approaches in Machine Learning and Cognitive Science: A Walkthrough: Volume 4
Автор: Vinit Kumar Gunjan, Jacek M. Zurada, Ninni Singh
Издательство: Springer
Серия: Studies in Computational Intelligence
Год: 2024
Страниц: 337
Язык: английский
Формат: pdf (true)
Размер: 11.8 MB

This book provides a systematic and comprehensive overview of cognitive intelligence and AI-enabled IoT ecosystem and Machine Learning, capable of recognizing the object pattern in complex and large data sets. A remarkable success has been experienced in the last decade by emulating the brain–computer interface. It presents the applied cognitive science methods and AI-enabled technologies that have played a vital role at the core of practical solutions for a wide scope of tasks between handheld apps and industrial process control, autonomous vehicles, IoT, intelligent learning environment, game theory, human computer interaction, environmental policies, life sciences, playing computer games, computational theory, and engineering development.

A cryptographic algorithm called SHA-256 converts data with variable lengths to data with fixed lengths [3]. The input is of variable length i.e.; “N” 512 blocks and each block is of size 512 bit.The “N” 512 bit input must be less than 264 bits. The resultant length of SHA-256 is of 256 bit. This 256 bit is called as a hash or message digests. The final hash acts as a digital signature in cryptographic security, and cryptocurrency. This paper proposes few optimization techniques in the architecture of SHA-256 hash algorithm namely addition of independent variables in the message compression function and addition of independent variables in the message scheduler function for improving hashing algorithm to minimize the critical path and to achieve less area utilization, power consumption and cells count that improves the performance of SHA-256 design.

The objective of steganography is to cover the actual presence of information exchange in unsuspecting digital media protections by embedding messages. The mechanisms of encryption, decryption and their usage in the protocols of communication are studied by cryptography or secret writing. In order to maintain security of information, both approaches transform data, although the principle of steganography is different than cryptography. Cryptography distorts an important message, but it does not hide the fact that there exists a message. The objective of cryptography is to render third party data unreadable, while steganography is intended to hide third party data. Both approaches come from an older year, although the contemporary field is rather young. The essential components of computer security are cryptography and Steganography. Cryptography is an established mathematical basis of computer safety and an area of computer science that is well developed and actively explored.

Food Detection with Image Processing Using Convolutional Neural Network (CNN): Using Deep Learning techniques and visual processing, the suggested method may identify the food. The network is trained using the Kaggle dataset. Existing technologies can process, assess, and recognise fruits and meals based on colour and texture cues. We were able to increase the adaptability and capabilities of the recognition system using AlexNet. A significant amount of data augmentation is required, in addition to learning the generalised pattern required to identify and recognise food items. The network separates the data into train data and test data categories. The accuracy is far higher than any other conventional models.

Google Appstore Data Classification Using ML Based Naive’s Bayes Algorithm: The Random Forest Algorithm, Tree Classifier based on Decision, Regression based on Logistics, Support Vector Machines, and KNN algorithm are only a few of the approaches for data categorization based on ML techniques that are described in this Paper. We classified the provided data using several methods, and by selecting various techniques, we will obtain various outcomes. While Naive’s Bayes is a classification technique for situations involving binary and many classes. When this method is explained using binary or categorical input values, it is easier to grasp. The mainstream of requests for this Naive Bayes method including analysis of sentiment, filtering spam and the systems recommendations, etc. This is simple and fast to put into practice, but the predictors’ dependence in most real-world scenarios makes the classifier less effective.