How vitormgou/fdipa-ann works
Project type and purpose
• This is a research-oriented analytical pipeline for a dissertation (“Algoritmo Dissertação”).
• It uses Jupyter notebooks to run experiments with variations of an algorithm (SGD, step‐size η, number of iterations, radius) and then produces CSV tables and plots.
File structure
Top-level items:
• FDIPA_com_DE.ipynb – main experiment notebook that implements the core algorithm
• filtragem_resultados.ipynb – notebook for filtering and post-processing raw results
• Untitled-1.ipynb – likely exploratory or auxiliary analysis
• plot_style.txt – shared plotting style definitions
Directories:
• TabelasSGD, TabelasVariacaoEta, TabelasVariacaoIteracao, TabelasVariacaoRaio
• each contains CSVs: acuracia_treino, acuracia_teste, norma_peso, perda_treino, perda_teste
• Plots_SGD, Plots_VariacaoEta, Plots_VariacaoIteracao, Plots_VariacaoRaio
• each will contain the charts generated from the corresponding tables
README for additional insights
• The README is minimal, only stating the high-level project name.
• We infer from file names that the focus is on comparing algorithmic hyperparameters and reporting training/test metrics.