B.M.S. College of Engineering

Abstract Large Language Models which consist of models like GPT5 from OpenAI, Claude Sonnet from Anthropic, Google’s Gemini are known for their reasoning capability, ability to summarize, cybersecurity analysis and other features. However, their reliability could be an issue particularly when dealing with ambiguous or hostile user input prompts. These kinds of attacks could lead to soft failures like false citations, hallucinations, and problematic suggestions. Current benchmarks (eg: TruthfulQA, AdvGLUE, and JailbreakBench) assess discrete aspects of robustness but are unable to identify multi-domain, conversational vulnerabilities that come up in practical applications. This paper introduces a novel dataset called MDH-Bench, a benchmark comprising 400 prompts that were specifically created to assess how reliably Large Language Models perform when they are provided with adversarial but natural and even tricky real-world queries. The benchmark covers eight types of challenges. They are numerical reasoning traps, security-sensitive defaults, historical and temporal inconsistencies, fictional research scenarios, contradictory context mixing, ethical provocations, and prompt-injection attempts. To monitor reliability in a holistic way, we present a Unified Multi-Domain Reliability Evaluation Framework that brings together several key metrics like accuracy, hallucination rate, hallucination degree (HD), unsafe output rate, and contradiction rate. A systematic assessment of four high-performance LLMs (GPT-5, Claude Sonnet 4, Gemini 2.5 Pro, DeepSeek V3.1) was conducted under controlled, manual, multi-annotator settings. The findings indicate that although overall accuracy is more than 88% in all models, hallucinations are still present both at high frequency and severity level. The results of our investigation showed that solely relying on accuracy is not enough for evaluating the reliability of a system. MDH-Bench and the combined reliability scoring system give a wide-ranging, adaptable base for the assessment and enhancement of trust in future LLMs for real-world applications.

Electric vehicles are the way forward as they have lesser carbon footprints and are more viable owing to the depleting fuel resources. The drives of electric vehicles present many advantages over traditional internal combustion engines, when considering emissions, efficiency, and no dependency upon oil. There has been a great advancement in the technology of batteries due to its development and the hybrid variant. This emphasizes the need to monitor the battery cells to prevent any occurrence of mishap which otherwise would lead to damages to property, life and personnel. This is achieved by utilizing a battery management system that allows for the monitoring of individual cells within a battery pack and offer protection against over-voltage, under-voltage, short circuit and indication of thermal runaway. Modern-day electric vehicles employ Li-ion batteries, which possess the merits of longer life cycles, flatter discharge curves and rapid charge-discharge capabilities, requiring utmost safety as they are more prone when compared to their lead-acid and nickel-cadmium counterparts.

MRI method is used in magnetic resonance imaging (MRI) to diagnose brain disorders by detecting the changing conditions within the brain. Peristaltic waves created on the flexible walls of the brain cause particles to move, and understanding this fluid flow can benefit the treatment of malignant tissues. In light of this, the electro-osmotic impact on the peristaltic flow of a fractional second grade fluid (ionic solution) in a microchannel with a heat and mass transfer is investigated. The governing equations of the modulated problem can be solved analytically by assuming a low Reynolds number (Re → 0) and a long wavelength (δ << 1). An analysis of the Poisson-Boltzmann equation is considered to examine the electro-kinetic mechanism. Using the Mathematica program, we solved a set of dimensionless equations and displayed the generated graphs. We analyzed relevant factors for their impacts on temperature, velocity, and pressure gradient. Biomedical engineers may find inspiration to create bio-microfluidic devices that aid in the movement of physiological fluids based on the model and its related findings.

The fake images and visuals can easily spread among social media users and they largely impact decisionmaking in society. Image forgery has become increasingly common as more non-professionals have access to image manipulation tools. These fake images are so sneaky that an ordinary person cannot guess them. Through social media, such photos are utilized to promote erroneous information in society. Image forgery detection is about segmenting the forged part from the images, primarily a region of interest. This paper suggests a unique method that depends on a dual attention network to detect forged segments. This network contains self-attention modules that contribute to extracting and matching features in the channel and spatial domains. These features help locate and identify the forged portions of digital images at various scales and channels. This experimental study uses typical datasets such as CASIA V1.0, CASIA V2.0, and Columbia. Proposed IFLNet technique outperforms other advanced techniques with a precision of 96 %, recall rate of 95 %, accuracy rate of 98 %, F1-score of 96 % and IoU score of 92 % for Columbia dataset and correspondingly other two datasets also.

In the current era, it is necessary to device authorization and authentication techniques to secure resources in information technology. There are several methods to substantiate authorization and authentication. User authentication is essential for authenticating user access control in WSNs. Biometric recognition error, lack of anonymity and vulnerability to attacks, user verification problem, revocation problem and disclosure of session key by the gateway node are some of the security flaws encountered. In this study, a Multimodal Authentication Scheme for Wireless Sensor Networks (WSN-MAS) is proposed to authenticate legitimate users. The main objective is the fusion of fingerprint and iris biometric features at feature level to enable additional accuracy to verify and match user identity with stored templates. In this paper, multimodal biometric features are used for authentication to improve performance, reduce system error rates to achieve better security in WSN.

Abrasive water jet machining possesses inherent technological and manufacturing advantages unmatched by most machine tools. They are generally used in automotive, marine and aerospace industries for machining of composites. However, abrasive water jet cutting of GFRP laminates possesses several challenges, such as surface finish and delamination of the cutspecimen. Henceforth, more experimental work is needed to provide sufficient machinability databases for manufacturing industries. This paper presents an experimental study for cutting GFRP laminates. Different AWJM conditions including pressure, traverse speed and abrasive flow rate are analyzed using full factorial design of experiments. Machining process responses such as surface roughness (Ra), top and bottom kerf width, kerf taper, delamination of fibers have been evaluated using analysis of variance (ANOVA) technique. The optimum process parameters for AWJM of glass fiber reinforced polymers have beensuggested.

Herein, we report selectively designed, synthesized and characterized bent-core molecules bearing methyl and chloro substituents at the lateral positions on the phenyl rings of the azobenzene wings for optical storage device applications.

In recent days as usage of gas is flexible and cheaper, it has been increased from home to hotels, industries, office etc. Besides its advantageous, gas leakage is becoming a major concern at all the places wherever it is being used as it may lead to a small accident or sometimes major disaster causing a great loss of infrastructures and lives of people. With the available advanced IoT and Mobile technologies, it is possible to build a system to prevent such accidents and informing the right person in case of tragedy. Hence, this paper has designed and developed a device which is capable of detecting the gas leakage and send alert message to specified mobile number. This proposed device could detect leakage of LPG, Methane, Butane or any such petroleum based gaseous substance using MQ5 Sensor (gas sensor) and send SMS to registered mobile number (by inserting sim in GSM module) and when the gas leakage is detected by the gas sensor then it sends an SMS alert to the registered mobile number thus notifying the owner to be careful and to take careful measures thus saving their lives and things. Implementation results of this device have proved its effectiveness.

Robotics is a current emerging technology in the field of science. A number of universities in world are working in this field. Robotics is the new emerging booming field, which will be of great use to society in the coming years. These days many types of wireless robots are being developed and are put to varied applications and uses.This paper concerns the design and fabrication of the Automated Guided Robot (AGR) prototype, as a mainframe in helping the robot to generate a self-understanding of the area of work and mobilization to a destination desired by the user. The main objective of this paper is to create and develop a Path Planning Mobile Robot able to avoid obstacles in its path and reach a target designated position from its starting point utilizing 4 wheel-based rover body, sensors, linear motors and microcontrollers. The paper describes the design process of the AGR development, from the initiation of idea to development of algorithms, to virtual prototype development of AGR mechanisms and controller board, to fabrication, to mobility, obstacle avoidance and path planning test. Two methods of path planning are presented, using compass bearing in undetermined space and predetermined path. Obstacle avoidance and path planning algorithms has been tested using the completed design of AGR body.

This paper is a case study of two different MOOC conducted in tandem with the academic curriculum. The paper projects on how the Project Based Learning helps to contribute to Outcomes-Based Education using MOOC as the medium of learning. The concepts of Agile methodologies like Pair Programming and Rapid Development can be inculcated into the student at an earlier stage to bring in industry ready engineers. We discuss how Android MOOC and the Blended MOOC contribute to Outcomes-Based Education.