Gonsalves Tad

The Traveling Salesman Problem (TSP) and the Vehicle Routing Problem (VRP) are two classical combinatorial optimization problems. In recent years, their drone-assisted variants, the Traveling Salesman Problem with Drones (TSP-D) and the Vehicle Routing Problem with Drones (VRP-D) have attracted growing attention. Generally, these problems are solved using exact algorithms or metaheuristic algorithms. However, as the problem complexity increases and the scale of instances grows, these approaches often become less efficient. In this paper, we propose a reinforcement learning method with a shared attention encoder and a hierarchical dual-decoder architecture, where truck–drone coordination is achieved by first decoding the truck’s next node and then conditionally decoding the drone action. To further explore the solution space of large-scale instances, the proposed method adopts a multi-rollout learning strategy. We conducted experiments on large-scale TSP-D and VRP-D instances, and the results show that this model outperforms traditional metaheuristic algorithms in terms of both solution quality and computational efficiency.

The two-tier vehicle routing problem (2T-VRP) represents a novel variant differing from the traditional VRPs. It can be applied to urban logistics operations system, offering significant potential for mitigating last-mile traffic congestion and reducing delivering costs. Distinct from traditional VRPs, this scenario contains a hierarchical two-tier structure, with trucks operating on the first tier and drones at stations on the second tier. In this study, we first extend the 2T-VRP framework by relaxing the hierarchical constraints, enabling trucks to transport goods not only to robot stations but also directly to customers. This new variant is referred to as the flexible two-tier vehicle routing problem with drone stations (F2T-VRP-DS). Then we formulate the problem as a mixed-integer linear programming (MILP) model. Finally, given the complexity of this problem, an improved adaptive large neighborhood search heuristic algorithm (IALNS) is proposed. The algorithm incorporates an adapted Clark and Wright saving algorithm as the initial solution, and a simulated annealing scheme is employed as the acceptance criterion. The experimental results show that our algorithm can provide high-quality solutions. In particular, compared with the MILP method, our algorithm demonstrates strong competitiveness on large-scale instances, offering smaller time consumption and better solution quality compared to commercial solver like Gurobi. In addition, based on the results of the sensitivity analysis, we further assessed the influence of several key parameters within the F2T-VRP-DS framework.

Automatic medical segmentation is crucial for assisting doctors in identifying disease regions effectively. As a state-of-the-art (SOTA) approach, generative AI models, particularly diffusion models, have surpassed GANs in generating high-quality images for tasks like segmentation. However, most diffusion-based architectures rely on U-Net designs with multiple residual blocks and convolutional layers, resulting in high computational costs and limited applicability on general-purpose devices. To solve this issue, we propose an enhanced denoising diffusion implicit model (DDIM) that incorporates lightweight depthwise convolution layers within residual networks and self-attention layers. This approach significantly reduces computational overhead while maintaining segmentation performance. We evaluated the proposed DDIM on two distinct medical imaging datasets: X-ray and skin lesion and polyp segmentation. Experimental results demonstrate that our model achieves, with reduced resource requirements, accuracy comparable to standard DDIMs in both visual representation and region-based scoring. The proposed lightweight DDIM offers a promising solution for medical segmentation tasks, enabling easier implementation on general-purpose devices without the need for expensive high-performance computing resources.

La rápida globalización y la creciente necesidad de comunicación interlingüística requieren corpus modernos y en tiempo real para ayudar a los estudiantes de idiomas. Los métodos tradicionales para crear dichos corpus, especialmente en español, son inadecuados debido a su incapacidad para procesar la gran cantidad de datos no estructurados disponibles en internet. En este estudio se exploran las metodologías de inteligencia artificial (IA) para la adquisición automática de documentos en español de la web, preprocesándolos y clasificándolos con el fin de construir un corpus vasto y flexible para el aprendizaje del español. La investigación aplica el rastreo web mediante el framework Scrapy para recopilar datos, que luego se limpian y clasifican utilizando modelos avanzados de procesamiento del lenguaje natural (PLN). En concreto, el estudio emplea el algoritmo BERT (Bidirectional Encoder Representations from Transformers) y su variante mejorada RoBERTa para lograr la clasificación de documentos. Mediante una combinación de técnicas de aumento de datos y modelos de aprendizaje profundo, el estudio logra una alta precisión en la clasificación de texto en español, lo que demuestra el potencial del uso de la IA para superar las limitaciones de los enfoques tradicionales de creación de corpus.

Resource Constraint Project Scheduling Problems with Discounted Cash Flows (RCPSPDC) focuses on maximizing the net present value by summing the discounted cash flows of project activities. An extension of this problem is the Payment at Event Occurrences (PEO) scheme, where the client makes multiple payments to the contractor upon completion of predefined activities, with additional final settlement at project completion. Numerous approximation methods such as metaheuristics have been proposed to solve this NP-hard problem. However, these methods suffer from parameter control and/or the computational cost of correcting infeasible solutions. Alternatively, approximate dynamic programming (ADP) sequentially generates a schedule based on strategies computed via Monte Carlo (MC) simulations. This saves the computations required for solution corrections, but its performance is highly dependent on its strategy. In this study, we propose the hybridization of ADP with three different metaheuristics to take advantage of their combined strengths, resulting in six different models. The Estimation of Distribution Algorithm (EDA) and Ant Colony Optimization (ACO) were used to recommend policies for ADP. A Discrete cCuckoo Search (DCS) further improved the schedules generated by ADP. Our experimental analysis performed on the j30, j60, and j90 datasets of PSPLIB has shown that ADP–DCS is better than ADP alone. Implementing the EDA and ACO as prioritization strategies for Monte Carlo simulations greatly improved the solutions with high statistical significance. In addition, models with the EDA showed better performance than those with ACO and random priority, especially when the number of events increased.

Jaringan saraf tiruan merupakan metode supervised learning yang telah diterapkan untuk menyelesaikan berbagai permasalahan klasifikasi. Sebagai metode supervised learning, jaringan saraf tiruan memerlukan data training untuk mengidentifikasi pola dalam data sehingga fase learning menjadi penting. Pada fase learning, konfigurasi bobot pada jaringan saraf tiruan diatur sehingga jaringan saraf tiruan tersebut bisa mengenali pola di dalam data. Pada penelitian ini diusulkan metode untuk mengoptimalkan nilai bobot pada konfigurasi jaringan saraf tiruan menggunakan pendekatan neuroevolution. Neuroevolution adalah pengintegrasian metode evolutionary algorithm; termasuk di dalamnya adalah berbagai metode metaheuristik; dengan  jaringan saraf tiruan. Secara khusus, penelitian ini menggunakan metode particle swarm optimization untuk mengoptimalkan bobot pada jaringan saraf tiruan. Kinerja model yang diusulkan dibandingkan dengan metode backpropagation dengan stochastic gradient descent menggunakan lima dataset: iris, wine, breast cancer, ecoli, dan wheat seeds. Hasil eksperimen menunjukkan bahwa model yang diusulkan memiliki akurasi yang lebih baik di tiga dataset dari lima dataset dan memiliki kinerja yang sama di dua dataset. Hasil penelitian ini mengindikasikan bahwa pendekatan neuroevolution memiliki potensi sebagai metode optimalisasi parameter pada jaringan saraf tiruan. Penelitian ini bisa dikembangkan dengan mengidentifikasi karakteristik konvergensi dari pendekatan neuroevolution maupun menerapkan berbagai metode evolutionary algorithm untuk mengoptimalkan nilai bobot pada jaringan saraf tiruan.

Background: Examination Timetabling Problem which tries to find an optimal examination schedule for schools, colleges, and universities, is a well-known NP-hard problem. This paper presents a Genetic Algorithm variant approach to solve a specific examination timetabling problem common in Japanese colleges and universities. Methods: The proposed algorithm uses direct chromosome representation Genetic Algorithm and implements constraint-based initialization and constraint-based crossover operations to satisfy the hard and soft constraints. An Island model with varying crossover and mutation probabilities and an improvement approach called pre-training are applied to the algorithm to further improve the result quality. Results: The proposed model is tested on synthetic as well as real datasets obtained from Sophia University, Japan and shows acceptable results. The algorithm was fine-tuned with different penalty points combinations and improvement combinations. Conclusion: The comparison results support the idea that the initial population pre-training and the island model are effective approaches to improve the result quality of the proposed model. Although the current island model used only four islands, incorporating greater number of islands, and some other diversity maintenance approaches such as memetic structures are expected to further improve the diversity and the result quality of the proposed algorithm on large scale problems.

<p>In this paper, the authors have proposed a computationally efficient, robust, and lightweight system for gait recognition. The proposed system contains two main stages: In the first stage, a classification network identifies optical flow corners in the normalized silhouette and calculates the distances traveled in every viewpoint which is further used by a regression model to identify the viewing angle. In the second stage, a feature extraction network computes the gait energy image (GEI) for every viewpoint and then uses principal component analysis (PCA) to extract low dimensional feature vectors from these GEI images. Finally, a multi-layer perceptron model is trained using the extracted principal components for every viewing angle. The performance of a system is comprehensively evaluated on the CASIA B and OULP gait dataset. The experimental results demonstrate the superior performance of a proposed system in viewing angle classification (100% accuracy), gait recognition (100% accuracy in normal walk), computational efficiency, robustness to clothing, and viewing angle variation.</p>

In CycleGAN, an image-to-image translation architecture was established without the use of paired datasets by employing both adversarial and cycle consistency loss. The success of CycleGAN was followed by numerous studies that proposed new translation models. For example, StarGAN works as a multi-domain translation model based on a single generator–discriminator pair, while U-GAT-IT aims to close the large face-to-anime translation gap by adapting its original normalization to the process. However, constructing robust and conditional translation models requires tradeoffs when the computational costs of training on graphic processing units (GPUs) are considered. This is because, if designers attempt to implement conditional models with complex convolutional neural network (CNN) layers and normalization functions, the GPUs will need to secure large amounts of memory when the model begins training. This study aims to resolve this tradeoff issue via the development of Multi-CartoonGAN, which is an improved CartoonGAN architecture that can output conditional translated images and adapt to large feature gap translations between the source and target domains. To accomplish this, Multi-CartoonGAN reduces the computational cost by using a pretrained VGGNet to calculate the consistency loss instead of reusing the generator. Additionally, we report on the development of the conditional adaptive layer-instance normalization (CAdaLIN) process for use with our model to make it robust to unique feature translations. We performed extensive experiments using Multi-CartoonGAN to translate real-world face images into three different artistic styles: portrait, anime, and caricature. An analysis of the visualized translated images and GPU computation comparison shows that our model is capable of performing translations with unique style features that follow the conditional inputs and at a reduced GPU computational cost during training.

Research on autonomous obstacle avoidance of drones has recently received widespread attention from researchers. Among them, an increasing number of researchers are using machine learning to train drones. These studies typically adopt supervised learning or reinforcement learning to train the networks. Supervised learning has a disadvantage in that it takes a significant amount of time to build the datasets, because it is difficult to cover the complex and changeable drone flight environment in a single dataset. Reinforcement learning can overcome this problem by using drones to learn data in the environment. However, the current research results based on reinforcement learning are mainly focused on discrete action spaces. In this way, the movement of drones lacks precision and has somewhat unnatural flying behavior. This study aims to use the soft-actor-critic algorithm to train a drone to perform autonomous obstacle avoidance in continuous action space using only the image data. The algorithm is trained and tested in a simulation environment built by Airsim. The results show that our algorithm enables the UAV to avoid obstacles in the training environment only by inputting the depth map. Moreover, it also has a higher obstacle avoidance rate in the reconfigured environment without retraining.

This paper presents a Genetic Algorithm approach to solve a specific examination timetabling problem which is common in Japanese Universities. The model is programmed in Excel VBA programming language, which can be run on the Microsoft Office Excel worksheets directly. The model uses direct chromosome representation. To satisfy hard and soft constraints, constraint-based initialization operation, constraint-based crossover operation and penalty points system are implemented. To further improve the result quality of the algorithm, this paper designed an improvement called initial population pre-training. The proposed model was tested by the real data from Sophia University, Tokyo, Japan. The model shows acceptable results, and the comparison of results proves that the initial population pre-training approach can improve the result quality.

Digital images often become corrupted by undesirable noise during the process of acquisition, compression, storage, and transmission. Although the kinds of digital noise are varied, current denoising studies focus on denoising only a single and specific kind of noise using a devoted deep-learning model. Lack of generalization is a major limitation of these models. They cannot be extended to filter image noises other than those for which they are designed. This study deals with the design and training of a generalized deep learning denoising model that can remove five different kinds of noise from any digital image: Gaussian noise, salt-and-pepper noise, clipped whites, clipped blacks, and camera shake. The denoising model is constructed on the standard segmentation U-Net architecture and has three variants—U-Net with Group Normalization, Residual U-Net, and Dense U-Net. The combination of adversarial and L1 norm loss function re-produces sharply denoised images and show performance improvement over the standard U-Net, Denoising Convolutional Neural Network (DnCNN), and Wide Interface Network (WIN5RB) denoising models.