RESUMO
In this paper, we present an analysis of important aspects that arise during the development of neural network applications. Our aim is to determine if the choice of library can impact the system's overall performance, either during training or design, and to extract a set of criteria that could be used to highlight the advantages and disadvantages of each library under consideration. To do so, we first extracted the previously mentioned aspects by comparing two of the most popular neural network libraries-PyTorch and TensorFlow-and then we performed an analysis on the obtained results, with the intent of determining if our initial hypothesis was correct. In the end, the results of the analysis are gathered, and an overall picture of what tasks are better suited for what library is presented.
Assuntos
Redes Neurais de ComputaçãoRESUMO
Myocardial infarction represents the most investigated pathology. Heart tissues are histologically assayed on a routine base in clinical laboratories. However, the lack of a standard operation procedure for e.g. calculating the size of the infarcted area of myocardium, may lead to an increased errors' interval, with little correlation between results of a same tissue and/or with other pathophysiological parameters. This creates a clear barrier for further development of novel therapeutic strategies. In the present study, we present the robust applicability of a novel methodology such as: advanced modular automated calculation of (i) the size of infarcted heart of mice, (ii) the net collagen content present in the scar, and (iii) the interstitial fibrosis in remote, which are simultaneously performed. The new approach of defining the infarct size, one of the important predictor of every cardiac therapeutic intervention, will create a positive impact in the research accuracy. Additionally, it will be expected that the readily assembled reports of simultaneously computed parameters and its user-friendly operation allow an efficient and effective estimation of measurements.
