Coverage for src/meta_learning/meta_learning_modules/utils_modules/configurations.py: 0%

1""" 

2Configuration Classes for Meta-Learning Utilities 

3================================================ 

4 

5Author: Benedict Chen (benedict@benedictchen.com) 

6 

7This module contains all configuration dataclasses and classes for meta-learning 

8utilities, providing type-safe configuration management with sensible defaults. 

9""" 

10 

11from dataclasses import dataclass 

12from typing import Optional, List, Dict, Any 

13 

14 

15@dataclass 

16class TaskConfiguration: 

17 """Configuration for meta-learning tasks.""" 

18 n_way: int = 5 

19 k_shot: int = 5 

20 q_query: int = 15 

21 num_tasks: int = 1000 

22 task_type: str = "classification" 

23 augmentation_strategy: str = "basic" # basic, advanced, none 

24 

25 # Configuration options for difficulty estimation methods 

26 difficulty_estimation_method: str = "pairwise_distance" # "pairwise_distance", "silhouette", "entropy", "knn" 

27 use_research_accurate_difficulty: bool = False # Enable research-backed methods 

28 

29 

30@dataclass 

31class EvaluationConfig: 

32 """Configuration for meta-learning evaluation.""" 

33 confidence_intervals: bool = True 

34 num_bootstrap_samples: int = 1000 

35 significance_level: float = 0.05 

36 track_adaptation_curve: bool = True 

37 compute_uncertainty: bool = True 

38 

39 # Configuration options for confidence interval methods 

40 ci_method: str = "bootstrap" # "bootstrap", "t_distribution", "meta_learning_standard", "bca_bootstrap" 

41 use_research_accurate_ci: bool = False # Enable research-backed CI methods 

42 num_episodes: int = 600 # Standard meta-learning evaluation protocol 

43 

44 # Additional configuration for advanced CI methods 

45 min_sample_size_for_bootstrap: int = 30 # Minimum sample size for bootstrap vs t-distribution 

46 auto_method_selection: bool = True # Automatically select best CI method based on data 

47 

48 

49class DatasetConfig: 

50 """Configuration for meta-learning dataset creation.""" 

51 

52 def __init__( 

53 self, 

54 dataset_type: str = "episodic", 

55 augmentation_strategy: str = "minimal", 

56 shuffle: bool = True, 

57 stratified: bool = True, 

58 normalize: bool = True, 

59 cache_episodes: bool = False, 

60 **kwargs 

61 ): 

62 self.dataset_type = dataset_type 

63 self.augmentation_strategy = augmentation_strategy 

64 self.shuffle = shuffle 

65 self.stratified = stratified 

66 self.normalize = normalize 

67 self.cache_episodes = cache_episodes 

68 for key, value in kwargs.items(): 

69 setattr(self, key, value) 

70 

71 

72class MetricsConfig: 

73 """Configuration for evaluation metrics computation.""" 

74 

75 def __init__( 

76 self, 

77 compute_accuracy: bool = True, 

78 compute_loss: bool = True, 

79 compute_adaptation_speed: bool = False, 

80 compute_uncertainty: bool = False, 

81 track_gradients: bool = False, 

82 save_predictions: bool = False, 

83 **kwargs 

84 ): 

85 self.compute_accuracy = compute_accuracy 

86 self.compute_loss = compute_loss 

87 self.compute_adaptation_speed = compute_adaptation_speed 

88 self.compute_uncertainty = compute_uncertainty 

89 self.track_gradients = track_gradients 

90 self.save_predictions = save_predictions 

91 for key, value in kwargs.items(): 

92 setattr(self, key, value) 

93 

94 

95class StatsConfig: 

96 """Configuration for statistical analysis.""" 

97 

98 def __init__( 

99 self, 

100 confidence_level: float = 0.95, 

101 num_bootstrap_samples: int = 1000, 

102 significance_test: str = "t_test", 

103 multiple_comparison_correction: str = "bonferroni", 

104 effect_size_method: str = "cohen_d", 

105 **kwargs 

106 ): 

107 self.confidence_level = confidence_level 

108 self.num_bootstrap_samples = num_bootstrap_samples 

109 self.significance_test = significance_test 

110 self.multiple_comparison_correction = multiple_comparison_correction 

111 self.effect_size_method = effect_size_method 

112 for key, value in kwargs.items(): 

113 setattr(self, key, value) 

114 

115 

116class CurriculumConfig: 

117 """Configuration for curriculum learning strategies.""" 

118 

119 def __init__( 

120 self, 

121 strategy: str = "difficulty_based", 

122 initial_difficulty: float = 0.3, 

123 difficulty_increment: float = 0.1, 

124 difficulty_threshold: float = 0.8, 

125 adaptation_patience: int = 5, 

126 **kwargs 

127 ): 

128 self.strategy = strategy 

129 self.initial_difficulty = initial_difficulty 

130 self.difficulty_increment = difficulty_increment 

131 self.difficulty_threshold = difficulty_threshold 

132 self.adaptation_patience = adaptation_patience 

133 for key, value in kwargs.items(): 

134 setattr(self, key, value) 

135 

136 

137class DiversityConfig: 

138 """Configuration for task diversity tracking.""" 

139 

140 def __init__( 

141 self, 

142 diversity_metric: str = "cosine_similarity", 

143 track_class_distribution: bool = True, 

144 track_feature_diversity: bool = True, 

145 diversity_threshold: float = 0.7, 

146 **kwargs 

147 ): 

148 self.diversity_metric = diversity_metric 

149 self.track_class_distribution = track_class_distribution 

150 self.track_feature_diversity = track_feature_diversity 

151 self.diversity_threshold = diversity_threshold 

152 for key, value in kwargs.items(): 

153 setattr(self, key, value)