Volume 14 | Issue 7 |

 Abstract

Automated thoracic disease screening from frontal chest X-rays is a multi-label learning problem where findings co-occur, supervision is often weak or uncertain, and deployment commonly involves domain shift and high interpretability requirements. NEPTUNE-CXR is a unified framework that couples anatomy-aware representation learning with uncertainty-aware decision outputs and clinically meaningful explanation artifacts. A frontal CXR is processed by APZ to estimate lung and heart masks, partition the lungs into six zones, and extract morphometrics; Z-ViT encodes a global token and zone tokens using mask-guided attention; CDN generates a closest-healthy counterfactual and residual evidence maps together with a distance-to-healthy plausibility signal; NSCG performs concept-graph reasoning to enforce cross-label consistency and produce concept activations and label logits; Uncertainty integrates evidential and Gaussian-process last-layer uncertainty with the plausibility signal to output calibrated multi-label probabilities. Outputs include per-label uncertainty scores, an abstention gate for selective prediction, conformal set-valued predictions, and explanation artifacts from zone attention, counterfactual/residual evidence, and concept activations. Evaluation is conducted on CheXpert (frontal-only) with comparisons against CuSeCXR and DEED, and optional transfer testing on one additional open-access corpus. The experimental study indicates improved discrimination, better calibration, stronger risk-coverage behavior, and more informative conformal sets, while providing evidence grounded in anatomy, counterfactual residuals, and concepts.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Chest X-Ray Screening; Multi-Label Classification; Zone-Aware Vision Transformer; Counterfactual Normalization; Neuro-Symbolic Concept Graph; Uncertainty Quantification; Selective Prediction; Conformal Prediction.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Some common dental diseases include caries, lesions, and structure disorders, among others. Such diseases require proper diagnosis to ensure their prevention in the future and treatment. Traditional ways of diagnosing them involve the use of specialist dentists who perform an examination based on the analysis of X-ray images. Nevertheless, such processes are inefficient and prone to errors made by specialists. Due to the rapid development of artificial intelligence, machine learning methods became a reliable way to conduct automated analysis of medical images.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Deep Lear.ning, Den.tal X-ray Anal.ysis, Convolu.tional Neu.ral Netw.orks (CNN), Medical Ima.ge Classif.ication, Grad.-CAM, Explai.nable Artif.icial Intell.igence (XAI), Dise.ase Detec.tion, Comp.uter Vis.ion, Healt.hcare Analy.tics, Ima.ge Proce.ssing

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Detection of brain tumors is a very significant area in medical diagnostics, since early detection can lead to better treatment outcomes and increased survival rate. MRI is one of the most common methods of brain diagnostics, as it gives detailed images of soft tissue. The manual analysis of MRI scans takes quite a lot of time and can suffer from inconsistencies caused by human factors. Therefore, this paper proposes a deep learning-based method of automated brain tumor detection and classification based on MRI images.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Brain Tum.or, MRI Ima.ges, Deep Lear.ning, Convolu.tional Neu.ral Netw.ork (CNN), Medi.cal Ima.ge Anal.ysis, Tum.or Detec.tion, Ima.ge Classif.ication, Compute.r- Aided Diagn.osis

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The increasing demand for electrical energy and the integration of renewable energy sources have created the need for intelligent power management systems. This paper presents a Smart Solar Grid with IoT-Based Load Shifting system designed to optimize energy utilization, reduce peak load demand, and improve grid efficiency. The proposed system integrates solar photovoltaic (PV) generation with Internet of Things (IoT) technology to monitor real-time energy consumption and automatically shift non-critical loads during peak demand periods.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Smart Grid, Solar Energy, Internet of Things (IoT), Load Shifting, Energy Management, Renewable Energy, Demand- Side Management.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The proposed Agro Mitra Robot presents an intelligent and automated solution for smart agriculture by integrating robotics, IoT, artificial intelligence, and image processing technologies for real-time monitoring, controlling, harvesting assistance, and plant disease detection. The system continuously monitors critical agricultural parameters such as soil moisture, temperature, and humidity, while autonomously performing tasks including irrigation control, pesticide spraying, and crop monitoring with minimal human intervention. A camera-based disease detection module using machine learning techniques enables early identification of plant infections, reducing crop loss and improving productivity. The robot also incorporates obstacle detection and autonomous navigation features to ensure efficient field operation. By minimizing labor dependency, optimizing resource utilization, and enhancing precision farming, the proposed system offers a cost-effective, sustainable, and efficient approach for modern agriculture, making it highly suitable for smart farming applications and future agricultural automation.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Smart Agriculture, Agro Mitra Robot, Internet of Things (IoT), Plant Disease Detection, Precision Farming, Artificial Intelligence, Image Processing, Autonomous Robot, Crop Monitoring, Automated Irrigation.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The project presents a hybrid renewable energy harvesting system designed to capture wasted ambient energy for smart city applications. It integrates two distinct energy sources: mechanical vibrations from utility poles and solar radiation. Piezoelectric sensors are employed to convert mechanical stress from pole vibrations into AC voltage, while a tilt-adjustable solar panel mechanism optimizes photon absorption by maintaining an ideal angle toward the sun. The system utilizes Maximum Power Point Tracking (MPPT) and a dedicated Battery Management System (BMS) to efficiently regulate and store the combined energy in a lithium battery. Delivering a reliable output of approximately 100-150 mW, this eco- friendly and scalable solution provides a consistent power supply for low-power IoT devices and remote infrastructure monitoring.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

DUAL SOURCE ENERGY RECOVERY FROM ELECTRIC POLE VIBRATION AND SOLAR PANEL TILT

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The increasing demand for renewable and sustainable energy sources has encouraged the development of smart energy harvesting systems. This paper presents a Hybrid Solar-Piezo Smart Energy Tile integrated with Local DC-DC Aggregation and an Intelligent Power Management System. The proposed model combines solar energy and piezoelectric energy harvesting techniques to generate electrical power efficiently from both sunlight and human footsteps. Solar panels convert solar radiation into electrical energy, while piezoelectric sensors generate electricity from mechanical pressure applied on the tile surface.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

HYBRID SOLAR PIEZO SMART ENERGY TILE WITH LOCAL DC-DC AGGREGATION AND INTELIGENT POWER MANAGER

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

College Introduction Robot is an innovative project that combines robotics, communications technology, and visual interaction. This Paper Presents the design and development of College Introduction Robot, a semi autonomous interactive system created to improve visitor engagement within educational institutions. Conventional campus guidance methods such as brochures, notice boards, and staff-led tours often suffer from limited availability, inconsistence explanations, and low interactivity. To address these challenges, the proposed robot delivers reliable campus information through voice-based communication, expressive visual feedback, and guided mobility. This system integrates Raspberry Pi 4 for high level processing and Arduino Mega for real time hardware control. Ultrasonic sensors support obstacle detection and safe navigation, while DC gear motors and motor drivers enable movement across campus environments. Voice recognition and question-answering functions allow the robot to respond naturally to visitor queries related to departments, facilities, admissions, and academic programs. OLED eye animations, TFT display messages, and handshake interaction using servo motors further enhance user engagement by creating a friendly and human-like communication experience. As a result, the system not only improves institutional presentation but also modernizes campus outreach by offering prospective students, parents, and academic visitors an engaging introduction to the college environment. An intelligent socially interactive campus robot enhances visitors engagement through autonomous guidance, responsive communication, and smart institutional assistance

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Arduino Mega, Embedded Systems, Human-Robot Interaction, Interactive Robotics, Raspberry Pi Voice Based Communication.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

A Hybrid Deep Learning Framework for Intelligent Phishing Website Detection

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

A Hybrid Deep Learning Framework for Intelligent Phishing Website Detection

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

As the global energy landscape pivots toward renew- able sources, the operational efficiency of wind farms has become a critical focal point for grid stability. The inherent stochasticity of wind resources necessitates a move from reactive to predic- tive control paradigms. This paper proposes a comprehensive four-layer architecture that harmonizes Programmable Logic Controllers (PLC), Supervisory Control and Data Acquisition (SCADA), Internet of Things (IoT) connectivity, and Machine Learning (ML). We validate this framework using the Berker ?i?mano?lu SCADA dataset, providing an empirical comparison between ensemble methods (Random Forest) and deep recurrent architectures (Long Short-Term Memory). Our findings indicate that both models achieve high predictive correlation (R2 > 0.96). Detailed analysis confirms that LSTM's ability to retain temporal state information via internal gates makes it more robust for tracking volatile power peaks. The integration of these layers facilitates a closed-loop optimization system capable of reduc- ing mechanical fatigue on turbine components and maximizing Annual Energy Production (AEP).

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

PLC, SCADA, IoT, Wind Farm, LSTM, Random Forest, Predictive Maintenance, MQTT, LoRaWAN, Industry 4.0, Renewable Energy.

  License

Creative Commons Attribution 4.0 and The Open Definition