Aditya Engineering College

Plastic Injection Molding (PIM) is key for making precise polymer parts. Shrinkage and warpage continue to be major issues. They lead to dimensional inaccuracies and part distortion, which reduce quality consistency. This affects the reliability of molded components. This study aims to develop a regression model to predict and enhance defects in polypropylene mobile covers. The study focused on important process parameters. These were melt and mold temperatures, packing and injection pressures, and cooling and packing times. It assessed how these factors influenced shrinkage and warpage. Scatter plots, histograms, and Q–Q plot analyses showed little bias and a nearly normal distribution of residuals. Feature importance analysis showed that cooling time (0.0834 s) and packing time (0.0724 s) are the main factors affecting shrinkage. Regression equations showed how parameters relate to defects. The Genetic Algorithm (GA) optimization reduced shrinkage to 1.42

The option in which QEDO directly has been in providing quantum-inspired methodologies with real-time systems in an enabling way to bring a paradigm shift in dynamic optimization in complex machine learning environments. QEDO implements some basic principles of quantum now via standard hardware, namely the concept of Superposition, entanglement and tunneling, via the so-called Variational Quantum Eigensolvers (VQE) and Quantum Annealing (QA), thus making it possible to explore high-dimensional search spaces significantly more efficiently than its classical analogs. Compared to modern non-adaptive QIO schemes (like QPSO or QAOA hybrids), which are hindered by changing constraints and latency greater than a second, QEDO employs a dynamic qubitentanglement mapping, which has enabled continuous model recalibration when using live data streams, with 32- fractional improvement in convergence and 94.96 fractional approach to accuracy during real-time provoked actions like object recognition, anomaly detection and predictive analytics. The accompanying framework overshadows prevailing baselines by 45 per cent in quality of solutions (p 0.001), consumes only 55 per cent of the CPU resources, and is easily scaled to more than 800 features undergoing attention (QAMA) and analytical dynamic analyses which are not truly scalable. With flexibility supporting both the deep learning, reinforcement learning, and probabilistic modeling paradigms of machine-learning effectively, and avoiding itself the heightening memory demands with sparsity-inducing representations, the hybrid architecture of QEDO is notably suitable to resource-constrained deployments, including autonomous robotics, financial trading, smart grids, and junction devices. Utilizing extensive empirical analyses, these benefits over classical optimizers and the latest quantumenhanced capabilities fill the conceptual divide between benefits of quantum-theory and practically available, operations-scale applications (e.g. in energy management, logistics and healthcare, to name a few). By introducing local minima avoidance by design through simulated non-local dynamics, QEDO is able to overcome local minima phenomena so prevalent in conventional gradient descent based optimization, which places it in a new status as a next-generation paradigm of using uncertainty-driven artificial intelligence.

This research presents a quick and effective method for Electrocardiogram (ECG) Signal Quality Assessment (SQA) using a Convolutional Neural Network (CNN). It utilizes the Fourier Magnitude Spectrum (FMS) to accurately differentiate between clean and noisy ECG signals. To ensure strong feature extraction, preprocessing was conducted via a Chebyshev Type II bandpass filter, which successfully removed out-of-band noise while retaining the key diagnostic features of the PQRST complex. The FMS was calculated using the Fast Fourier Transform (FFT), providing a thorough spectral analysis. A streamlined 1-D CNN model was developed and fine-tuned for devices with limited resources, achieving outstanding classification results with $\mathbf{1 0 0 \%}$ accuracy, $\mathbf{9 9. 3 \%}$ sensitivity, and $\mathbf{9 5. 4 \%}$ specificity. In comparison to conventional techniques, the proposed method showcased enhanced efficiency, lower computational demands, and suitability for real-time applications. These findings underscore the model's potential for integration into portable and embedded ECG monitoring systems, aiding in precise and prompt cardiac health evaluations.

The present research suggests an innovative smart home automation system using NLP and IoT technology, providing efficient control of household appliances through voice and text instructions. This system combines an NLP module with the Home module, offering natural communication between humans and devices without the need for pre-defined command set. The NLP module takes up the user commands by applying speech recognition, tokenization, intent classification, and entity extraction algorithms, whereas the Home module controls the appliances via Arduino UNO with the use of relays. Real time control of light, fan, and door in household environment becomes possible through text or voice communication. Experiments show that this approach is efficient in increasing the user convenience and accessibility, while decreasing dependence on complicated interfaces. The proposed intelligent automated control system is cost-effective to future intelligent smart home systems.

The diagnosis of skin diseases has been a focus of great interest because of the increased rates of skin disorders and the necessity of prompt and convenient diagnosis of dermatological disorders. In this paper, we introduce a machine-learning model of the multi-class classification of nine common skin diseases with the help of the customized Convolutional Neural Network (CNN). Eight hundred and seventy-eight dermoscopic images were acquired at Kaggle, processed, and augmented followed by classification using the proposed CNN architecture. The images were downsized to 200 × 200 pixels, and rescaled, rotated, sheared, zoomed and horizontally flipped. The model attained a training accuracy of 92.78