Module synthetic 

Synthetic Data Generation for AutoML.

This module provides automatic synthetic data generation capabilities to improve model performance in low-resource domains. Generated data is validated, quality-scored, and integrated into the AutoML optimization loop.

Quick Start

use aprender::synthetic::{SyntheticConfig, GenerationStrategy};

// Configure synthetic data generation
let config = SyntheticConfig::default()
    .with_augmentation_ratio(0.5)
    .with_quality_threshold(0.7);

assert_eq!(config.augmentation_ratio, 0.5);
assert_eq!(config.quality_threshold, 0.7);

Design Principles

• Quality-First: All generated samples validated before inclusion
• Diversity-Aware: Monitors for mode collapse and distribution shift
• AutoML Integration: Augmentation parameters jointly optimized with model hyperparameters

References

• Cubuk et al. (2019). AutoAugment: Learning Augmentation Strategies. CVPR.
• Wei & Zou (2019). EDA: Easy Data Augmentation. EMNLP.
• Ratner et al. (2017). Snorkel: Weak Supervision. VLDB.

Re-exports

pub use andon::AndonConfig;

pub use andon::AndonEvent;

pub use andon::AndonHandler;

pub use andon::AndonSeverity;

pub use andon::DefaultAndon;

pub use andon::TestAndon;

Modules

andon

Andon mechanism for synthetic data generation (Toyota Jidoka).

cache

Caching for Synthetic Data Generation.

code_eda

Code-Specific EDA (Easy Data Augmentation) for source code.

code_features

Code Feature Extraction for Commit-Level Analysis.

eda

Easy Data Augmentation (EDA) for text data.

mixup

MixUp Data Augmentation.

shell

Shell Autocomplete Synthetic Data Generator.

template

Template-based synthetic data generation.

weak_supervision

Weak Supervision for Synthetic Data Generation.

Structs

DiversityMonitor

Monitors diversity of generated synthetic samples over time.

DiversityScore

Diversity metrics for a batch of generated samples.

QualityDegradationDetector

Detects when synthetic data is hurting rather than helping model performance.

SyntheticConfig

Configuration for synthetic data generation.

SyntheticStream

Streaming iterator for memory-constrained synthetic generation.

SyntheticValidator

Validates generated synthetic samples before inclusion.

Enums

GenerationStrategy

Available synthetic data generation strategies.

SyntheticParam

Synthetic data generation hyperparameters.

ValidationResult

Result of validating a synthetic sample.

Traits

SyntheticCallback

Callback trait for monitoring synthetic data generation.

SyntheticGenerator

Trait for synthetic data generators.

Functions

check_andon

Check Andon conditions and trigger events if thresholds exceeded.

generate_batched

Generate synthetic data in batches to manage memory.