Journal of Artificial Intelligence, Machine Learning and Soft Computing

ABSTRACT

Artificial Intelligence (AI) systems are rapidly becoming part of critical domains such as healthcare, finance, transportation, governance, and defense. While these systems provide significant advantages in automation and decision making, concerns about safety, reliability, robustness, and trustworthiness have also increased. AI failures can result in serious consequences, including biased decisions, system crashes, adversarial exploitation, and unintended behavior. This paper presents a comprehensive review of AI safety, robustness, and verification techniques that aim to ensure reliable performance of AI systems in real-world conditions. It discusses challenges associated with adversarial attacks, distribution shifts, model uncertainty, and interpretability. Further, the paper reviews verification and validation approaches including formal methods, testing strategies, explainability tools, and runtime monitoring. Practical frameworks, evaluation metrics, and recent research trends are summarized to guide researchers and practitioners in designing dependable AI systems. Tables and figures are provided for better understanding of techniques and comparisons.

KEYWORDS: AI Safety, Robustness, Verification, Adversarial Attacks, Explainable AI, Formal Methods, Trustworthy AI, Model Validation

ABSTRACT

The rapid urbanization across the globe has intensified challenges related to infrastructure, traffic management, energy consumption, and public safety. Artificial Intelligence (AI) has emerged as a transformative tool for designing and managing smart cities and urban systems, enabling real-time decision-making, predictive analytics, and enhanced resource allocation. This paper presents a comprehensive review of AI applications in smart cities, covering traffic and transportation systems, energy management, public safety, environmental monitoring, and citizen-centric services. Furthermore, we explore key AI techniques employed in urban systems, including machine learning, deep learning, and reinforcement learning. The paper also discusses challenges, such as data privacy, interoperability, and scalability, and outlines future research directions for AI-enabled urban development.

KEYWORDS: Smart cities, Artificial Intelligence, Urban systems, Machine learning, Deep learning, IoT, Traffic management, Energy optimization

ABSTRACT

Artificial Intelligence (AI) has become a fundamental technology in modern robotics and autonomous systems. From industrial manipulators to self-driving vehicles and service robots, AI enables machines to perceive environment, make decisions, and act intelligently with minimal human control. The combination of machine learning, computer vision, sensor fusion, and control algorithms allows robots to perform complex tasks in uncertain and dynamic environments. This paper reviews the major role of AI techniques in robotics, including perception, planning, navigation, learning, and human-robot interaction. It also discusses applications in manufacturing, healthcare, agriculture, military, and domestic assistance. Challenges such as safety, energy efficiency, computational limits, and ethical issues are also examined. The study presents recent trends and future directions of AI-enabled robotics systems.

KEYWORDS: Artificial Intelligence, Robotics, Autonomous Systems, Machine Learning, Computer Vision, Path Planning, Human-Robot Interaction

Abstract

Artificial Intelligence (AI) has emerged as a transformative technology in healthcare, particularly in diagnosis and prognostics. AI-powered systems can analyze complex medical datasets, identify patterns in patient data, and predict disease outcomes with high accuracy. This review explores AI methodologies applied in healthcare, including machine learning (ML), deep learning (DL), natural language processing (NLP), and hybrid approaches. The paper highlights AI applications in radiology, pathology, genomics, and clinical decision support systems. Furthermore, challenges such as data privacy, algorithmic bias, interpretability, and clinical integration are discussed. Finally, future directions are outlined, emphasizing AI’s role in personalized medicine and predictive healthcare.

Keywords: Artificial Intelligence, Healthcare, Diagnosis, Prognostics, Machine Learning, Deep Learning, Predictive Analytics, Clinical Decision Support

Abstract

In this paper, we are presenting a proposed system for Smart Surveillance model in which we are implementing facial monitoring system by embedded face detection and face tracking algorithm and it consists of three steps namely: facial detection, feature extraction and recognition by using Haar classifier and Eigen face approach. The purpose of the project is to make a system, which would detect and take snapshots and videos of the Human motion when detected and upload to an external server.

Keywords: Face detection, Face recognition, Face authentication, Raspberry Pi, IoT

Abstract

Machine Learning is an Approach to AI that uses a system that is capable of learning from experience. It is intended not only for AI goals (e.g., copying human behavior) but it can also reduce the efforts and/or time spent for both simple and difficult tasks like stock price prediction. In other words, ML is a system that can recognize patterns by using examples rather than by programming them. Big data analytics in security involves the ability to gather massive amounts of digital information to analyze, picture and draw insights that can make it possible to predict and stop cyber attacks. Along with security technologies, it gives us stronger cyber defense posture. They allow organizations to recognize patterns of activity that represent network threats. In this paper, I emphasis on how Big Data be able to progress information security best practices. I am trying to apply machine learning procedures in cyber security using big data Analytics.

Keywords: Machine Learning Big Data, Cyber Security, Artificial Intelligence, Deep Learning.

Abstract

This paper chronicles the development of an artificial neural network designed to recognize handwritten digits. The neural network described here is not a general-purpose neural network, and it's not some kind of a neural network workbench. Rather, we will focus on one very specific neural network (a five-layer convolutional neural network) built for one very specific purpose (to recognize handwritten digits). One of our goals here was to reproduce the accuracy achieved by Dr. LeCun, who was able to train his neural network to achieve 99.18% accuracy (i.e., an error rate of only 0.82%). This error rate served as a type of "benchmark", guiding our work.

Keywords:  Digit Recognition, Neural Networks, Handwritten Digit Recognition, Artificial Neural Networks

Abstract

Real estate properties assessment is the price estimation process for real estate properties. Nowadays, real estate brokers provide easy access to detailed online information on real estate properties to their clients. Regularly, the repeat sales model has been widely adopted to estimate real estate price. Generally for price prediction Regression is used i.e. prediction of continuous valued-function. But here we are going to use Deep Neural Network in order to improve productivity and accuracy. We introduce a deep learning approach to smartly and effectively evaluating real estate values. We propose a systematic method to derive a layered knowledge graph and design a structured Deep Neural Network (DNN) based on it. Neurons in a structured DNN are structurally connected, which makes the network time and space efficient; and thus, it requires fewer data points for training. The DNN model has been planned to learn from the most recently taken data points. We propose a systematic method to derive a layered knowledge graph and design a structured Deep Neural Network based on it. We introduce a deep learning approach to smartly and effectively assessing real estate values.

Keywords: Deep Neural Network (DNN), layered knowledge graph, structurally connected, deep learning approach.

Abstract

REAL estate appraisal, which is the price estimation process for real estate properties, for both buyers and sellers it is essential as the basis for compromise and deal. Customarily, the repeat sales model has been widely adopted to estimate real estate price. Though, it depends on the design and calculation of a complex economic related index, which is challenging to estimate accurately the estate’s price. Nowadays, real estate brokers provide easy access to detailed online information on real estate properties to their clients. In a smart city, effective and accurate real estate assessments governed by a local government is crucial for determining the property taxes. Such assessments have never been trivial, and inappropriate assessments may result in disputes between property owners and the local government. Generally for price prediction Regression is used (Prediction of continuous valued-function). But here we are going to use Structured Deep Neural Network in order to improve efficiency and accuracy. We introduce a deep learning approach to smartly and effectively assessing real estate values. We propose a systematic method to derive a layered knowledge graph and design a structured Deep Neural Network (DNN) based on it. Neurons in a structured DNN are structurally connected, which makes the network time and space efficient; and thus, it requires fewer data points for training. The structured DNN model has been designed to learn from the most recently captured data points; therefore, it allows the model to adapt to the latest market trends.

Keywords: Deep Neural Network (DNN), Structured Deep Neural Network, layered knowledge graph, structurally connected, deep learning approach.

Abstract

This article proposes a computer learning system. This system allows an effective conversation between the primary teacher and the student. Based on theories of psychology and theories of artificial psychology, knowledge of learning has been developed. The meaning emotion fundamental theory is that four basic emotions and four types of teaching psychology are defined and the bi-dimensional is based on the theory of dimension emotion. This system gives the teacher mental status and psychological value. Finally, this system was supposed to be used to identify the basis of digital image processing technology.

Keywords: Artificial Psychology, Affective Computing, Image Processing, Pattern Recognition, Q-processing, Facial Expression Recognition, B/S structure, Artificial emotion, Intelligent interaction, Psychological Assistive Technology, Aging service.

Abstract

This Paper will involve developing a model that generates suitable captions for images. This will help in analyzing image and converting the textual content to other useful forms. Image descriptions provide textual information about non-text content that appears on your website, allowing it to be presented auditory, as visual text, or in any other form that is best for the user. Image descriptions are plain text descriptions of images, gifs, videos, and other media. Humans have been captioning images involuntary since decades and now in the age of social media where every image has a caption over various social platforms. Psychologically those things are affected by events and scenarios running in mind or influenced by nearby activities and emotion. Sometimes those are far-far away from real context. Describing the content of an image is a fundamental problem in artificial intelligence that connects computer vision and natural language processing.

Keywords: Image Captioning, Artificial Intelligence, Python

Abstract

Crying is the primary sound child makes when he/she enters the world. It is a response to address different circumstances including pain, hungry, lonely, discomfort etc. Parents or guardians endeavour to recognize and timely address the reason for the infant cry before crying sets in. Anyway first time they often fail. This prompts to dissatisfaction and feeling of helplessness. This paper proposes the system to predict the reason for the baby cries using machine learning. To predict the causes, this system takes baby crying audio as input, processes it, gives causes and suggestions about the actions to be taken, as output.

Keywords: Machine learning, infant cry, MFCC, KNN, python_speech_features.

Abstract

In this project we have presented an approach to generate questions from a paragraph and the size of the paragraph is defined by its scope. A mix of syntax and semantic based approach to natural language processing is used to generate the questions from the paragraph. Important sentences from the paragraph are selected based upon the certain features and the questions are generated for these selected sentences. Our system implements generation of question from paragraph and also generating simple and complex types of questions. And the research till date works on either implementing question generation from single sentences or implementing generation of simple questions from paragraph or implementing question generation of complex questions from paragraph.

Keywords: Question Generation, Tokenization, POS, NLP, Data mining, etc 

Abstract

Artificial Intelligence (AI) systems are increasingly deployed in healthcare, finance, governance, transportation, education, and many other domains. While AI offers tremendous benefits in automation and decision-making, it also raises serious ethical concerns related to bias, privacy, accountability, transparency, and misuse. Ethical and responsible AI systems aim to ensure that AI technologies are developed and deployed in ways that respect human values, fairness, and societal well-being. This paper presents a comprehensive review of ethical challenges in AI and discusses frameworks, principles, and technical approaches to build responsible AI systems. Topics such as fairness, explainability, accountability, data governance, human oversight, and regulatory efforts are examined. The paper also presents practical methods, tools, and case studies for implementing ethical AI. Finally, challenges and future directions for responsible AI development are discussed.

Keywords: Ethical AI, Responsible AI, AI fairness, Explainable AI, AI governance, Bias mitigation, AI accountability, Human-centered AI

Abstract

Breast cancer is the second largest cause of cancer deaths among women. At the same time, it is also among the most curable cancer types if it can be diagnosed early. Research efforts have reported with increasing confirmation that the support vector machines (SVM) have greater accurate diagnosis ability. In this paper, breast cancer diagnosis is based on a SVM-based method combined with Random Forest Classifier has been proposed. Experiments have been conducted on different training-test partitions of the Wisconsin breast cancer dataset (WBCD), which is commonly used among researchers who use machine learning methods for breast cancer diagnosis. The performance of the method is evaluated using classification accuracy, correlation matrix, scattering matrix. The result shows that the highest classification accuracy (100%) is obtained for the random forest classifier model that contains five features, and this is very promising compared to the previously reported results

Keywords:- Support vector machine, Random Forest Classifier, Wisconsin breast cancer (WBCD), machine learning, correlation matrix, scattering matrix

Authors: Apoorva Shanbhag, Deekshith G Nayak, Pai Akash Anilkumar*, Saniya, Narayan Naik

Abstract: The main idea behind this project is to detect and classify Brain Tumor using MRI images. Brain Tumor is an abnormal intracranial growth caused by cells reproducing themselves in an uncontrolled manner. The algorithm incorporates steps for pre-processing feature extraction and image classification using neural network techniques. Human investigation is the routine technique for brain MRI tumor detection and tumor classification. Interpretation of images is based on organized and explicit classification of brain MRI and also various techniques have been proposed. Information identified with anatomical structures and potential abnormal tissues which are noteworthy to treat are given by brain tumor segmentation on MRI, the proposed system uses the adaptive pillar KMeans algorithm for successful segmentation and the classification methodology is done by the two-tier classification approach. The proposed two-tier classification system classifies the brain tumors in double training process which gives preferable performance over the traditional classification method. The proposed system has been validated with the support of real data sets and the experimental results showed enhanced performance.

Abstract

Robot is a machine that is specially programmed by a user or computer. Robot is mainly used for carrying things with complex action and it can be guided by external control device. This article depicts that controlling of robot is based on hand gesture movement. The gesture movement is done with the help of 3-axis accelerometer (i.e.ADXL 335).It works based on tilting the accelerometer in various directions. Based on the accelerometer directions the robot is controlled. An accelerometer is placed on the hand that is used to transmit the signal that is configured with Arduino UNO and the controlling the motor is done with NI-MYRIO using LabVIEW. Based on the transmitting device value the robot movement is controlled in various directions like forward, backward, left, right and stop. The movement of indication is viewed in LabVIEW.

Keywords: 3-axis Accelerometer, Arduino UNO, LabVIEW, NI-myRIO

Abstract

The autonomous car and unmanned ground vehicle is a vehicle that is capable of sensing its environment and navigating without human input. Some believe that autonomous vehicles have the potential to transform the transportation industry and cleaning up the environment. Levels of Autonomous car: No-Automation (Level 0) - The driver (human) controls it all. Function specific automation (Level 1) - Some control functions such as the electronic stability control or charged brakes is automated. Combined function automation (Level 2) - At least two main control functions such as the adaptive cruise control in combination with lane centering are automated. Limited self-driving automation (Level 3) - Under certain traffic and environmental conditions, the driver cedes full control of all safety–critical functions and rely heavily on the vehicle to watch for any changes in the conditions requiring transition to driver control. Full selfdriving automation (Level 4) vehicle is intelligently designed to monitor roadway conditions and act solo and performing all safety–critical driving functions for an entire trip (a fully driverless level). The working of an Autonomous car can be viewed in three stages; they are: The sensing unit of an autonomous car consists of various sensors such as Lidar, Infrared, Cameras, etc. The signals from the sensing unit are sourced to the Logical processing unit, which is responsible for the decision making, user interface, etc. The Mechanical Control System is the unit which regulates the metrics of car. The most important aspect of any autonomous car is the Artificial Intelligence that drives it, it is the element that replaces the human factor.

Keywords: Machine Learning, Convolutional Neural Networks, Deep Learning.

Abstract

Transfer Learning and Domain Adaptation have emerged as powerful techniques in Artificial Intelligence that reduce the need for large labeled datasets and improve model performance across different but related tasks. Traditional machine learning models require extensive data and training time when applied to new problems. However, transfer learning allows knowledge gained from one task to be reused for another task, while domain adaptation deals with changes in data distribution between source and target domains. These approaches are widely used in computer vision, natural language processing, healthcare, speech recognition, and autonomous systems. This paper presents a comprehensive review of transfer learning and domain adaptation, discussing their types, methodologies, architectures, applications, challenges, and recent advancements. Comparative tables and illustrative figures are included to better explain various techniques. The study also highlights practical considerations and future research directions in this rapidly growing field.

Keywords: Transfer Learning, Domain Adaptation, Knowledge Reuse, Deep Learning, Source Domain, Target Domain, Feature Extraction, Model Fine-tuning

Abstract

With the advances in data technology (IT) criminals ar victimization Internet to commit varied cyber crimes. Cyber infrastructures ar extremely susceptible to intrusions and alternative threats. Physical devices and human intervention aren't comfortable for observation and protection of those infrastructures; therefore, there's a requirement for additional refined cyber defense systems that require to be versatile, labile and strong, and able to find a good type of threats and create intelligent time period selections. Varied bio-inspired computing strategies of AI are progressively taking part in a vital role in cyber crime detection and interference. The aim of this study is to gift advances created to date within the field of applying AI techniques for combating cyber crimes, to demonstrate however these techniques may be an efficient tool for detection and interference of cyber attacks, similarly on provide the scope for future work.

Keywords: Cyber Crime, Artificial Intelligence, Intelligent Cyber Defense Methods, Intrusion Detection and Prevention Systems, Computational Intelligence

Abstract

Worldwide, around 10% of the population has dyslexia, a specific learning disorder. Most of previous eye tracking experiments with people with and without dyslexia have found differences between populations suggesting that eye movements reflect the difficulties of individuals with dyslexia. In this paper, we present the first statistical model to predict readers with and without dyslexia using eye tracking measures. The model is trained and evaluated in a 10-fold cross experiment with a dataset composed of 1,135 readings of people with and without dyslexia that were recorded with an eye tracker. Our model, based on a Support Vector Ma- chine binary classifier, reaches 80.18% accuracy using the most informative features. To the best of our knowledge, this is the first time that eye tracking measures are used to predict automatically readers with dyslexia using machine learning.

Categories and Subject Descriptors K.4.2 [Computers and Society]: Social Issues—Assistive technologies for persons with disabilities; I.2.1 [Artificial Intelligence]: Applications and Expert Systems—Medicine and science.

Keywords: Dyslexia, eye tracking, eye movements, diagnosis, detection, prediction, machine learning, support vector machine

Abstract

The era of mobile technology opens the windows to the mobile applications. Along with these applications, the use of intelligent computer programs are observable. It is related to similar task of using machines to understand human intelligence. Combining both the usage of mobile applications with use of Artificial Intelligence is increasing at an alarming rate One application that falls into category is the “Image Recogniser and Captioner†developed for Android phones. The prime objective of the “Image Recogniser and Captioner†is to create a full fledge Android application which could allow users to search for anything from mobile phones by just clicking the pictures. It recognises the content of the image along with the body posture the human being. The result is obtained in the caption describing about the image.'

Keywords: Image Recogniser and Captioner”, Artificial Intelligence (AI)

Abstract

In machine learning, a convolutional neural network (CNN, or ConvNet) is a class of deep, feed-forward artificial neural networks that has successfully been applied to analyzing visual imagery. This paper, proposes the possibility of learning deep network structures and the relative study of their wide range of applications. CNNs use a variation of multilayer perceptron designed to require minimal preprocessing. The engineering of CNN is accomplished with neighborhood associations and tied weights overtaken by some type of pooling which brings about interpretation invariant highlights. Another advantage of CNNs is that they are less demanding to prepare and have numerous less parameters than completely associated systems with a similar number of concealed units. An Artificial Neural Network (ANN) with numerous concealed layers between the info and yield layers is a profound neural system (DNN). Deep learning structures of CNN, have been connected to fields of computer vision, speech recognition, natural language processing, audio recognition, social network filtering, machine translation and bioinformatics where the delivered comes about tantamount to and at times better than human specialists.

Keywords: Neural Network (CNN), Field-Programmable Gate Array (FPGA), Deep Neural Networks, Drop-out algorithm.

Abstract

Artificial Intelligence (AI) has shown potential to address many societal problems and support public policy making. This paper reviews how AI applications contribute to social good across different sectors, outlines challenges and risk, and discusses how public policy frameworks can guide responsible AI deployment. We explore domains such as healthcare, education, urban planning, disaster management, and public governance. While AI has benefits, it also brings issues on bias, privacy and fairness. This paper argues that effective public policy must balance innovation with ethical safeguards, encourage data governance and include multiple stakeholders in design and regulation of AI systems. We conclude with policy recommendations that aims to maximize positive societal impacts while managing risk.

Keywords: AI for social good, public policy, ethics, bias, data governance, civic technology, public governance

Abstract

Traditional machine learning models depend heavily on large labeled datasets for achieving good performance. However, in many real-world situations, collecting large amounts of labeled data is expensive, time consuming, or even impossible. This limitation motivated the development of Zero-Shot Learning (ZSL) and Few-Shot Learning (FSL) techniques, which aim to recognize new classes with little or no training examples. These approaches try to mimic the human ability of learning new concepts from very limited experience. Zero-shot learning relies on auxiliary information such as semantic attributes or textual descriptions to identify unseen classes, while few-shot learning leverages a small number of examples with advanced training strategies like meta-learning and metric learning. This paper presents a detailed review of the principles, techniques, architectures, and applications of ZSL and FSL. Recent advancements using deep learning, transformers, and large language models are also discussed. Challenges, evaluation methods, and future research directions are highlighted.

Keywords: Zero-Shot Learning, Few-Shot Learning, Meta-Learning, Transfer Learning, Metric Learning, Deep Learning, Semantic Embedding