Web caching is essential for the World Wide Web, saving processing power, bandwidth, and reducing latency. Many proxy caching solutions focus on buffering data from the main server, neglecting cacheable information meant for server writes. Existing systems addressing this issue are often intrusive, requiring modifications to the main application for integration. We identify opportunities for enhancement in conventional caching proxies. This paper explores, designs, and implements a potential prototype for such an application. Our focus is on harnessing a faster bulk-data-write approach compared to single-data-write within the context of relational databases. If a (upload) request matches a specified cacheable URL, then the data will be extracted and buffered on the local disk for later bulk-write. In contrast with already existing caching proxies, Squid, for example, in a similar uploading scenario, the request would simply get redirected, leaving out potential gains such as minimized processing power, lower server load, and bandwidth. After prototyping and testing the suggested application against Squid, concerning data uploads with 1, 100, 1.000, ..., and 100.000 requests, we consistently observed query execution improvements ranging from 5 to 9 times. This enhancement was achieved through buffering and bulk-writing the data, the extent of which depended on the specific test conditions.

This paper focuses on the hypothesis of optimizing time series predictions using fractal interpolation techniques. In general, the accuracy of machine learning model predictions is closely related to the quality and quantitative aspects of the data used, following the principle of garbage-in, garbage-out. In order to quantitatively and qualitatively augment datasets, one of the most prevalent concerns of data scientists is to generate synthetic data, which should follow as closely as possible the actual pattern of the original data. This study proposes three different data augmentation strategies based on fractal interpolation, namely the Closest Hurst Strategy, Closest Values Strategy and Formula Strategy. To validate the strategies, we used four public datasets from the literature, as well as a private dataset obtained from meteorological records in the city of Braşov, Romania. The prediction results obtained with the LSTM model using the presented interpolation strategies showed a significant accuracy improvement compared to the raw datasets, thus providing a possible answer to practical problems in the field of remote sensing and sensor sensitivity. Moreover, our methodologies answer some optimization-related open questions for the fractal interpolation step using Optuna framework.

With the availability of several remotely sensed data sources, the problem of efficiently visualizing the information from multisource data for improved Earth observation becomes an intriguing and challenging subject. Multispectral and hyperspectral images encompass a wealth of spectral data that standard RGB monitors cannot replicate directly. Thus, it is important to elaborate methods for accurately representing this information on conventional displays. These images, with tens to hundreds of spectral bands, contain relevant data about specific wavelengths that RGB channels cannot capture. Traditional visualization methods often use only a limited amount of the available spectral information, resulting in a significant loss of information. However, recent advances in artificial intelligence models have provided superior visualization techniques. These AI-based methods allow for more realistic and visually appealing representations, which are important for the information interpretation and direct identification of areas of interest. The main goal of our study is to process aggregated datasets from various sources using a fully connected neural network (FCNN), while considering visualization as a secondary objective. Given that our data come from a variety of sources, a significant emphasis in our study was placed on the preprocessing stage. In order to achieve a consistent visualization across datasets from different sources, proper preprocessing by standardization or normalization procedures is essential. Our research comprises numerous experiments to demonstrate the effectiveness of the proposed technique for image visualization.

With the increasing demands for safe, robust and high efficient systems, the performance indicators should be quantified over the system’s life-cycle. Condition monitoring drastically reduce the maintenance costs, prevents unscheduled working interruption of the system, keeps productivity performance and the system in safe operating mode. Artificial intelligence and machine learning models are able to process big data sets, harness the data and predict the failures. This paper presents a method to monitor the states conditions and to identify the faults that may appear during the operation of an industrial elevator by developing a Long Short-Term Memory model.

One of the most discussed topics at present is related to optimization processes. We are living in a world in which everything gets more and more expensive every day. So, to avoid a significant crisis, we need to re-evaluate the cost of everything that surrounds us, including the materials used in construction. The problem that comes is not related to the materials themselves, but the quantities used. For each structure, we can deduce two requirements: practicability and resistance. To meet these, the tendency is to create a structure with more support than it needs just to guarantee safety. Unfortunately, this implies a greater weight that the structure must support and also greater costs for building it. The object of this paper is to properly develop a Genetic Algorithm based on the methods of mathematical programming for this class of structural optimization problems in the presence of multiple design constraints. This state-of-the-art heuristic solution, alongside other related works, should demonstrate the utility of Genetic Algorithms in these types of problems, and improve the time necessary for the implementation of an optimal architecture.

Current trends involve machine learning techniques, Artificial Intelligence, etc. in most Industry4.0 specific research directions. In the automotive industry, but not only, methods and new machine learning algorithms appear in order to mainly shorten the development time of new components and their validation. The aim of this paper is to generate an input set of data, starting from a classic system existing in the Simcenter Amesim platform, use it as input data in a machine learning analysis and validate the new proposed machine learning methodology. This approach seeks to analyze a vehicle suspension model by using an artificial neural network. Essential for this work are the data sets on which the neural network is trained, as these require an exceptional degree of accuracy and robustness for the result to be as close as possible to mathematical calculations. The final aim is to help create a model enabling the prediction of the values of the vehicle’s suspension travel, speed and acceleration.

Bearing failure is one of the foremost causes of breakdown in rotating machinery. Such failures can be catastrophic and can result in costly downtime. Thus, several methods it has been proven to identify the signature of bearing faults and it was proved for medium to high speed regimes that it is possible to detect a fault through vibration signals. A previous research [1] includes a comparative analysis of the benefits of applying the different methods of machine learning for obtaining key statistical features in the time domain to characterize a fault. The paper reviews various models, using the same dataset and commenting on the various results obtained. Some final conclusions and supporting experiments related to Random Forest and Decision Tree models have been made, emphasizing the importance of each model in the case of the chosen study.