Professor Celestine Iwendi

The creation of digital marketing has enabled companies to adopt personalized item recommendations for their customers. This process keeps them ahead of the competition. One of the techniques used in item recommendation is known as item-based recommendation system or item-item collaborative filtering. Presently, item recommendation is based completely on ratings like 1-5, which is not included in the comment section. In this context, users or customers express their feelings and thoughts about products or services. This paper proposes a machine learning model system where 0, 2, 4 are used to rate products. 0 is negative, 2 is neutral, 4 is positive. This will be in addition to the existing review system that takes care of the users' reviews and comments, without disrupting it. We have implemented this model by using Keras, Pandas and Sci-kit Learning libraries to run the internal work. The proposed approach improved prediction with 79% accuracy for Yelp datasets of businesses across 11 metropolitan areas in four countries, along with a mean absolute error (MAE) of 21%, precision at 79%, recall at 80% and F1-Score at 79%. Our model shows scalability advantage and how organizations can revolutionize their recommender systems to attract possible customers and increase patronage. Also, the proposed similarity algorithm was compared to conventional algorithms to estimate its performance and accuracy in terms of its root mean square error (RMSE), precision and recall. Results of this experiment indicate that the similarity recommendation algorithm performs better than the conventional algorithm and enhances recommendation accuracy.

Accurate stock price forecasting remains a challenging yet crucial task in the financial industry due to the non-linear relationships, noisy, and time-dependent nature of the market data. This study presents a deep learning approach known as long-short-term memory (LSTM) for predicting the closing prices of stock using historical data. The model is designed to capture the complex temporal dependencies inherent in stock market sequences, addressing the limitations of traditional statistical models such as ARIMA and linear regression. Using key key characteristics such as past closing prices, the LSTM model achieved high predictive performance with a Mean Squared Error (MSE) of 0.00036, a mean absolute error (MAE) of 0.0096, and a coefficient of determination (R²) of 0.9941, indicating strong generalization and accuracy. The results demonstrate the effectiveness of LSTM architectures in time series forecasting for financial applications. This research contributes to the development of robust and automated decision support tools for investors and sets a performance benchmark for future deep learning models in stock market prediction.

The rapid advancement in cloud computing technology is continually evolving, with threat actors refining their tactics, exploiting new vulnerabilities, and expanding their influence. This dynamic environment exposes cloud infrastructure to emerging cyber-attacks, including Advanced Persistent Threats (APT), impacting both customers and service providers. Understanding the gap in APT detection literature is crucial for researchers. The research aims to comprehensively understand APTs' influence on cloud security, analyse existing approaches, emulate adversary plans, simulate attacks using Mitre Caldera, employ Snort for detection, and utilise the Nessus vulnerability scanning tool. The study addresses critical questions about APTs' exploitation of cloud environments, strengths and weaknesses of mitigation methods, impacts of successful APT attacks, vulnerabilities in cloud infrastructures, and techniques for detecting APTs. The findings underscore the intricate interplay between APT activities and cloud environments, emphasising the need for robust detection and mitigation strategies. The combination of APT simulation, vulnerability assessment, and detection mechanism analysis yields invaluable insights into the evolving threat landscape within cloud ecosystems. As organisations increasingly embrace cloud technologies, the lessons from this study contribute substantially to the ongoing discourse on fortifying cloud security against persistent and evolving cyber threats.

Autistic adolescents often experience the inability to identify their emotions and self-regulate them, thus creating the impulse for the construction of intelligent assistive technologies. Building on this premise, this work proposes a novel Psycho-Intelligent Dialogue Agent (PIDA) system, which attempts to incorporate advances in affective computing, contextualized reasoning , and psychotherapeutic dialogues, in aiding emotional self-regulation with teenagers with autism spectrum disorder (ASD). This system integrates a visual emotion recognition model with an adaptive conversational bow. To train the emotion classifier for real time application trained using transfer learning techniques based on the VGG16 architecture of deep convolutional neural networks, it was trained on a specialized dataset comprising of autistic children's facial expressions and achieved an accuracy of 71% at a 5-emotion recognition task. The Effect recognition module serves the context-aware dialogue manager in real time adapting and personalizing the emotional regulation frameworks to be employed. PIDA's dialogues are based on the principles of clinical psychotherapy, with psychotherapeutic techniques and intervention strategies which are individually tuned to the emotional state and contextual parameters of the situation. The system was designed and built salted with caregiver integration features to enable guardians to monitor progress and active participant in the personalization of the intervention. Primary experimental results reflect the feasibility of this dimension in emotional awareness and emotional regulation and coping strategies. To support we provide uninterrupted emotional assistance to autistic young people and offer flexible support resources during and in between emotional therapy appointments.

Autistic Adolescents usually experience difficulties in the management of emotions, routine transitions and social cue interpretation. Many existing tools that aim to fill in the gap are often non-personalise, static or lack real-time responsiveness in handling these challenges. This study conceptualises and empirically validates a prototype of an emotionally adaptive AI companion that focuses on reducing stress due to routine transition, emotional regulation and social cue interpretation while increasing personalised management by providing contextual support. A quasi-experimental, mixed methods design is adopted. The core of this system conducts facial multimodal emotion recognition through facial expression and simulated voice tone using transfer learning across three CNN architectures (ResNet-18, MobileNetV2, and EfficientNet-B0) as comparison tests. The resulting emotion output is feeds into a contextual engine for real-time personalised interventions which can also be continuously improved through critical feedback-in-the-loop control architecture based on caregiver logs. The key model trade-offs are validated, the findings established that ResNet-18 possesses the highest accuracy of 48%, EfficientNet-B0 with a superior F1 Score of 0.31 and MobileNetV2 proves to be efficient but slightly lower performance compared to other architectures. Simulated user feedback validation resulted in high preliminary acceptability, as high as 87.5% acceptability for an intervention like ”Reassurance”. This validated the utility of this responsive system. This transfer-learning based, multi-modal pipeline is robust. The results of the comparative analysis uncovered a very profound and instructive trade-off between the complexity of models, their efficiency, and performance metrics relating to accuracy versus the F1-score

Bank card fraud is one of the biggest challenges in digital finance space, which needs detection models to address class imbalance, interpretability, and adaptability to changing tactics of fraud. The paper proposes a neuro-adaptive architecture established on a highly structured preprocessing pipeline with stratified splitting, feature normalisation, and representation learning via a Denoising Autoencoder. At the core of this framework lays an Artificial Neural Network optimised by the Firefly Algorithm for fast hyperparameter tuning facilitated by Elastic Weight Consolidation that promotes continual learning without sacrificing past performance. The proposed Adaptive ANN + FA outperforms baseline ANN, CNN, and LSTM models mainly in F1-Score, precision, and recall-the main metrics in fraud detection. Also, SHAP breaks out feature contribution and prediction reasonability making the results very transparent. Optimised adaptive and explainable models are positioned here as strong enablers of real-world fraud discovery and subsequent robustness in the financial systems.

The fast-growing convergence of neuroscience, behaviour computing, and adaptive artificial intelligence (AI) offers the possibility to transform human, machine interaction. This work presents Psycho-Intelligence, a new, closed-loop system that merges electroencephalography (EEG) and inertial motion unit (IMU) signals to adaptively recognise and react to users' cognitive and affective states. Levying low-cost wearable sensors (Muse EEG and MPU-6050), the system has real-time signal acquisition , sophisticated preprocessing, spectral and statistical feature extraction, as well as multimodal fusion features. Dimensionality reduction and feature selection techniques, including Principal Component Analysis and XGBoost gain metrics, enhance learning optimally. Multiple machine learning algorithms like Random Forest, SVM and XGBoost are trained to identify engagement states with high accuracy, warranted by extensive testing through cross-validation, ROC AUC, and F1-scores. The pipeline is incorporated into an adaptive feedback system that can regulate chatbot tone, learning material, or interactive graphics based on detected user states. Statistical validation with linear mixed models confirms the robustness of EEG-derived measurements in engagement prediction. The research establishes a new paradigm for emotionally intelligent AI systems and provides a technical foundation for ethical, real-time psycho-behavioural intelligence for communication networks, education systems, and cognitive health monitoring.

This study investigates the risk of job automation in London due to artificial intelligence (AI), applying supervised machine learning techniques to identify occupations most at risk. Leveraging a dataset encompassing job-specific features such as primary tasks, industry domains, and associated AI models, the research develops two predictive models. A Random Forest Classifier is used to categorize jobs as low, medium, or high automation risk, while a Linear Regression model estimates the proportion of each occupation's workload likely to be automated. The Random Forest model achieved a high accuracy rate of 97 % in classifying job risk, indicating strong predictive capability. Meanwhile, the regression model explained 85 % of the variance in the AI workload ratio, highlighting a significant relationship between job attributes and automation potential. These results suggest that job characteristics are reliable indicators of AI impact, particularly in routine, repetitive, and low-skilled roles that are more easily codified and replicated by algorithms. The findings align with broader economic theories such as creative destruction and technological waves, suggesting that AI not only displaces certain roles but also drives structural transformation within the labor market. By focusing on London, this study provides a localized understanding of how AI is reshaping employment patterns. It underscores the growing urgency for strategic workforce re-skilling and adaptive policy frameworks to mitigate negative outcomes and maximize opportunities presented by AI. Ultimately, this research contributes valuable insights into the interaction between AI technologies and employment, helping policymakers, employers, and educators anticipate change and prepare for a more resilient, inclusive labor market.

Inclusion in recruitment remains a critical challenge in the evolving landscape of work, particularly for individuals with neurodivergent cognitive profiles. This study investigates the potential of adaptive Artificial Intelligence (AI) assessments to foster more inclusive hiring practices for neurodiverse talent. Using thematic analysis and simulation-driven data, the research explores how AI systems can be optimized to support equity, reduce cognitive bias, and enhance candidate-job alignment. Grounded in Adaptive Theory, Evidence-Based Rationale, and the Input-Environment-Output (I-E-O) framework, the study models an AI-powered recruitment pipeline using psychometric clustering and retention data. Results reveal that while AI tools can personalise assessments and improve hiring outcomes, they also risk amplifying bias if not calibrated across diverse cognitive profiles. The study emphasizes the importance of neurodiversity-aware model tuning, participatory design involving neurodivergent individuals, and the integration of fairness metrics such as demographic parity and equal opportunity. The findings contribute practical insights and a replicable framework for advancing inclusive hiring through AI-enabled systems.

The impart of climatic change apparently affect accessibility of good water for agriculture irrigation system in addition to human dependency and demand on farm products for survival, increases water unavailability challenges in the farm affecting ecosystem when there is no balance between country food production capacity and population growth. Agricultural sector is challenged with unequal distribution of water in the farm plantations reducing food production strength, this is more severe in under-developing regions whereby require smart irrigation systems as promising solution to mitigate against threat of water distribution to farm products. This research work design and develop smart irrigation system with less expensive micro-controller components, implementing water distribution system using sensor captured information of soil condition, infusing precision irrigation system that calculate and supplies exact water require based on soil dryness level. The research adopts internet of things IoT technology where sensor will be used to monitor and capture soil information and control water distribution based on available soil dataset. The outcome of the research gives absolute control on over-irrigation and under-irrigation system that increases agricultural productions with advance technological means, precision irrigation system mechanisms reduces water wastage and ensure equal distributions. Multivariate soil type test will greatly enhance general acceptability of this concept in cross-functional regions.