Tasriqul Islam

Most AI Governance Professionals Are Missing the Point : Here’s What Really Causes Damage and How to Fix It
Linkedin
September 18, 2025

To the AI and data governance community: I see you. The effort and expertise many of you bring to managing AI risks is impressive. But I have to be blunt...too many governance frameworks today are set up for failure.

FUNDAMENTALS OF MACHINE LEARNING TECHNIQUES
Xoffencerpublication
September 12, 2025

Machine learning is a subfield of computing science that evolved both from the knowledge obtained through the process of learning how to classify data based on that understanding and also from the understanding gained through the process of learning the computational-based concepts of Artificial Intelligence, or AI. Machine learning, also known as ML, is a common abbreviation for the field. To put it another way, machine learning is the process of training computers to learn on their own via their interactions with data without being explicitly taught to do so. This is accomplished through the use of artificial neural networks. Both humans and animals may claim to be the first to conceptualize what we now call learning. There are a lot of similarities to be discovered between the way that machines learn and the way animals learn. In point of fact, many of the methods that are now used in machine learning were first created to imitate the foundations of animal and human learning using computer representations. This was done to further the field of artificial intelligence. The basic scientific concept of habituation, for instance, outlines the process by which an animal progressively ceases reacting to a stimulus that has been repeatedly shown to the animal. If a dog is taught to perform a range of tasks, such as rolling over, sitting, picking up objects, etc., it is considered to be an outstanding example of animal learning since it is capable of considerable learning if it is trained to do so. If a dog is taught to execute a number of tasks, such as rolling over, sitting, picking up items, etc., it is considered to be an excellent example of animal learning. Many people believe that dogs are the best representatives of animal intelligence. As opposed to the preceding example of successful learning, there aren't many real world applications of machine learning that we can point to as evidence that it's a helpful notion in the current world. This is in contrast to the earlier demonstration of successful learning. Virtual personal assistants, traffic predictions using GPS navigation, surveillance of multiple cameras by AI to detect crime or unusual behavior of people, social media uses ML for face recognition and news feed personalization, search engine result refinement, e-mail spam filtering where a machine memorize all the previously labeled spam e-mails by the user, and a lot more applications are just some of the many places where ML is widely used. Other applications include: a lot more applications. By using all of these applications, it has become abundantly evident that making use of knowledge and experience that one already has will result in a more efficient learning process. The close link that ML has to computational statistics, which also plays a vital role, makes the process of making predictions more simpler and more straightforward. Everyone is entitled to wonder "why does a machine need to learn something?" and there is no wrong answer to this question. There are just a few compelling arguments in favor of the need of machine learning. The fact that we just said that the development of learning capabilities in robots may help us better understand how animals and people gain information should not come as a surprise to anybody.

EEG Signal Based Human-Computer Interaction based Autonomous Driving Control System
IEEE
April 09, 2025

EEG signal analysis and driving intention prediction are the main challenges of human-centric autonomous intelligent driving control systems. This paper uses spectral characteristics and EEG functional brain networks to recognise straight-line, left-turn, and right-turn driving. The method classifies and identifies driving processes using support vector machines and Gaussian mixture models. Simulated driving results show that the suggested technique accurately identifies many driving activities. Out of 16 individuals, straight-line and turning recognition accuracy is regularly above 82%, peaking at 86.66%. The recognition accuracy for left- and right- and turn motions peaks at 77.95%. Examining critical area dependency during turning processes shows lateralisation differences. Specifically, left turns require more inter-brain region activity than right turns.