本科毕设基于DAG区块链的联邦学习框架实现去中心化和个性化源码+使用说明.zip

import matplotlib.pyplot as plt import numpy as np import pandas as pd import ast import os from collections import Counter import csv from scipy.interpolate import make_interp_spline, BSpline from sknetwork.clustering import Louvain, modularity from .node import Node class Graph: def __init__(self, data, generation, analysis_output_dir=None): self.nodes = [Node(n["id"], n["parents"], n["metadata"]) for n in data["nodes"] if n["metadata"]["time"] <= generation] self.generation = generation self.analysis_output_dir = analysis_output_dir def print_statistics(self, include_reference_statistics=True): #### Helper methods for printing def _print(text): if self.analysis_output_dir: analysis_filepath = os.path.join(self.analysis_output_dir, "statistics.txt") with open(analysis_filepath, "a+") as statistics_file: print(text, file=statistics_file) else: print(text) def _print_multiple_statistics_lines(labels, data, text): for l, d in zip(labels, data): _print(text % (l, d)) _print("") #### Get statistics data and print it statistics = self._get_statistics_data(include_reference_statistics) _print(f"Average accuracy in the last 5 rounds: {statistics['average_accuracy_last_5_rounds']}") _print('') _print("Average clients per round: %f" % statistics["average_clients_per_round"]) _print("") _print("Average parents per round (not including round 1): %f" % statistics["average_parents_per_round"]) _print("") # Pureness _print_multiple_statistics_lines( *statistics["average_pureness_per_round_approvals"], "Average pureness (approvals) for %s per round: %f") if (include_reference_statistics): _print_multiple_statistics_lines( *statistics["average_pureness_per_round_ref_tx"], "Average pureness (ref_tx) for %s per round: %f") # Information gain _print_multiple_statistics_lines( *statistics["information_gain_per_round_approvals"], "Average information gain (approvals) for %s per round: %f") if (include_reference_statistics): _print_multiple_statistics_lines( *statistics["information_gain_per_round_ref_tx"], "Average information gain (ref_tx) for %s per round: %f") def plot_transactions_per_round(self, smooth_line=False, plot_axis_labels=True, plot_for_paper=False): data = self._get_num_transactions_per_round() self._line_plot( title='Number of transactions per round', data_arrays=[data], y_label="Number of transactions", smooth_line=smooth_line, plot_axis_labels=plot_axis_labels, plot_for_paper=plot_for_paper) def plot_parents_per_round(self, plot_first_round=False, smooth_line=False, plot_axis_labels=True, plot_for_paper=False): data = self._get_mean_parents_per_round(plot_first_round) self._line_plot( title='Mean number of parents per round (%s round 1)' % ("including" if plot_first_round else "excluding"), data_arrays=[data], y_label="Mean number of parents", smooth_line=smooth_line, plot_axis_labels=plot_axis_labels, plot_for_paper=plot_for_paper) def plot_accuracy_boxplot(self, print_avg_acc=False, plot_axis_labels=True, plot_for_paper=False, ymax=1.0, ylabel=""): data = self._prepare_acc_data() plt.boxplot(data) if print_avg_acc: plt.plot([i for i in range(1, self.generation + 1)], [np.mean(x) for x in data]) # Settings for plot plt.title('Accuracy per round') # Fix y axis data range to [0, 1] plt.ylim([0, ymax]) if plot_axis_labels: plt.xlabel("Round") plt.xticks([i for i in range(1, self.generation + 1)], [i if i % 10 == 0 else '' for i in range(1, self.generation + 1)]) plt.ylabel(ylabel) def save_or_plot_fig(format="png"): if self.analysis_output_dir: analysis_filepath = os.path.join(self.analysis_output_dir, f"accuracy_per_round.{format}") plt.savefig(analysis_filepath, format=format) else: plt.show() save_or_plot_fig() if plot_for_paper: # Remove title plt.title("") save_or_plot_fig(format="pdf") # Clear canvas for next diagram plt.clf() def plot_information_gain_ref_tx(self, smooth_line=False, plot_axis_labels=True, plot_for_paper=False): labels, data_arrays = self._get_information_gain_ref_tx() # Plot data self._line_plot( title='Information gain (reference tx)', data_arrays=data_arrays, labels=labels, smooth_line=smooth_line, plot_axis_labels=plot_axis_labels, plot_for_paper=plot_for_paper) def plot_information_gain_approvals(self, smooth_line=False, plot_axis_labels=True, plot_for_paper=False): labels, data_arrays = self._get_information_gain_approvals() # Plot data self._line_plot( title='Information gain (approvals)', data_arrays=data_arrays, labels=labels, smooth_line=smooth_line, plot_axis_labels=plot_axis_labels, plot_for_paper=plot_for_paper) def plot_pureness_ref_tx(self, smooth_line=False, plot_axis_labels=True, plot_for_paper=False): labels, data_arrays = self._prepare_reference_pureness(compare_to_ref_tx=True) self._line_plot( title='Cluster pureness (reference transaction)', data_arrays=data_arrays, labels=labels, smooth_line=smooth_line, plot_axis_labels=plot_axis_labels, plot_for_paper=plot_for_paper) def plot_pureness_approvals(self, smooth_line=False, plot_axis_labels=True, plot_for_paper=False): labels, data_arrays = self._prepare_reference_pureness() self._line_plot( title='Cluster pureness (approvals)', data_arrays=data_arrays, labels=labels, smooth_line=smooth_line, plot_axis_labels=plot_axis_labels, plot_for_paper=plot_for_paper) def plot_avg_age_difference_ref_tx(self, smooth_line=False, plot_axis_labels=True, plot_for_paper=False): avg_age_difference_per_round = self._prepare_data_avg_age_difference_to_ref_tx() self._line_plot( title='Average age difference to reference transaction per round', data_arrays=[avg_age_difference_per_round], y_label='Age in rounds', smooth_line=smooth_line, plot_axis_labels=plot_axis_labels, plot_for_paper=plot_for_paper) def plot_modularity_per_round(self, smooth_line=False, plot_axis_labels=True, plot_for_paper=False): m = self._prepare_modularity() self._line_plot( title='Modularity per round', data_arrays=[[x for x, _, _, _ in m]], y_label='modularity', smooth_line=smooth_line, plot_axis_labels=plot_axis_labels, plot_for_paper=plot_for_paper) def plot_num_modules_per_round(self, smooth_line=False, plot_axis_labels=True, plot_for_paper=False): m = self._prepare_modularity() self._line_plot( title='Modules per round', data_arrays=[[x for _, x, _, _ in m]], y_label='#modules', smooth_line=smooth_line, plot_axis_labels=plot_axis_labels, plot_for_paper=plot_for_paper) def plot_misclassification_per_round(se