Update README.md

README.md

@@ -5,6 +5,8 @@ tags:
 - image-classification
 - data-augmentation
 - small-sample-learning
+- empirical-study
+- rethinking
 datasets:
 - cifar100
 - cifar10
@@ -12,52 +14,53 @@ metrics:
 - accuracy
 ---
 
-# Prior-Guided Augmentation: Experiment Results
-# 先验引导增强：实验结果备份库
+# Rethinking Augmentation: Experiment Results
+# 小样本增强策略再思考：实验结果库
 
-Official checkpoint and experimental results repository for the paper: 
-**"Prior-Guided Augmentation: A Reliable Strategy for Small-Sample Datasets"**
+Official experiment results repository for the paper:  
+**"When More is Not Better: Rethinking Data Augmentation under Small-Sample Regimes"**
 
-This repository contains the official model weights, search logs, and visualizations for the Prior-Guided Augmentation (PGA) framework.
-本仓库包含该论文配套的官方模型权重、搜索日志以及可视化图表。
+**TL;DR**: We find that in small-sample regimes (e.g., CIFAR-100 100-shot), complex augmentation strategies (like RandAugment) often yield diminishing returns and high instability. A single, well-tuned operation can achieve comparable accuracy with significantly lower variance.
+
+**核心发现**：在小样本场景下，盲目增加数据增强的复杂度（如 RandAugment）往往收益递减且带来极大的不稳定性。我们发现，单一且经过调优的增强操作能在保持精度的同时显著降低方差。
 
 ---
 
-## 📊 Core Results / 核心实验结果
+## 📊 Key Findings / 核心发现
 
-### CIFAR-100 (100-shot)
-| Metric / 指标 | Result / 结果 |
-| :--- | :--- |
-| **Top-1 Accuracy** | 40.74% |
-| **Stability (Std)** | 0.78 (Highest in benchmarks) |
-| **Optimal Policy** | ColorJitter (m=9, p=0.5) |
+### 1. Stability-Accuracy Trade-off / 稳定性与精度的权衡 (CIFAR-100 100-shot)
+| Policy / 策略 | Val Acc (%) | Stability (Std) | Note |
+| :--- | :--- | :--- | :--- |
+| **RandAugment** (N=2,M=9) | 42.24% | 1.17 | High Variance (Unstable) |
+| **Single-Op** (Ours) | 40.74% | **0.78** (Lowest) | **Stable & Reliable** |
+| **Baseline** | 39.90% | 1.01 | - |
 
-### CIFAR-10 (50-shot)
+### 2. Zero-Variance Generalization / 零方差泛化 (CIFAR-10 50-shot)
 | Metric / 指标 | Result / 结果 |
 | :--- | :--- |
 | **Top-1 Accuracy** | 50.0% |
-| **Stability (Std)** | 0.0 (Zero Variance / 零方差) |
+| **Stability (Std)** | **0.0** (Zero Variance across 3 seeds) |
 
 ---
 
 ## 📂 Repository Structure / 仓库结构
 
-- `checkpoints/`: Model weights (`best_model.pth`). / 官方模型权重。
-- `figures/`: Visualizations (Heatmaps, Efficiency plots). / 论文可视化图表（热图、效率对比图等）。
-- [phase_c_final_policy.json](phase_c_final_policy.json): The discovered optimal augmentation policy. / 搜索出的最优增强策略。
-- [cifar10_50shot_results.csv](cifar10_50shot_results.csv): CIFAR-10 experimental data. / CIFAR-10 50-shot 实验原始数据。
-- [stability_seeds_results.csv](stability_seeds_results.csv): Raw data for stability tests across random seeds. / 跨随机种子的稳定性原始测试数据。
-- [destructiveness_metrics.csv](destructiveness_metrics.csv): LPIPS/SSIM analysis for semantic preservation. / 语义保真度分析（LPIPS/SSIM）。
+- `checkpoints/`: PyTorch model weights (`best_model.pth`). / 官方模型权重。
+- `figures/`: Paper visualizations (Heatmaps, Trade-off plots). / 论文可视化图表（热图、权衡对比图等）。
+- [phase_c_final_policy.json](phase_c_final_policy.json): The discovered optimal single-operation policy. / 搜索出的最优单一增强策略。
+- [cifar10_50shot_results.csv](cifar10_50shot_results.csv): CIFAR-10 generalization experiment data. / CIFAR-10 泛化实验数据。
+- [stability_seeds_results.csv](stability_seeds_results.csv): Raw data verifying the stability claim. / 验证“稳定性”结论的原始数据。
+- [destructiveness_metrics.csv](destructiveness_metrics.csv): LPIPS/SSIM analysis for semantic preservation. / 语义保真度分析数据。
 
 ---
 
 ## 📜 Citation / 引用
-If you find this repository useful, please cite our work:
-如果您觉得本工作对您有帮助，请引用我们的工作：
+If you find this study useful, please cite our work:
+如果您觉得这项研究对您有启发，请引用我们的工作：
 
-(Stay tuned! Citation will be updated upon acceptance. / 论文接收后将在此更新引用信息。)
+*(Citation will be updated upon acceptance / 论文接收后更新)*
 
 ---
 
 ## 🔗 Links / 相关链接
-- **GitHub**: [imnotnoahhh/prior-guided-aug-policy](https://github.com/imnotnoahhh/prior-guided-aug-policy)
+- **Code & Paper**: [imnotnoahhh/Rethinking-Augmentation](https://github.com/imnotnoahhh/Rethinking-Augmentation)