Update README.md

README.md

@@ -12,8 +12,8 @@ base_model:
 
 # 🔍 **AntiDeepfake**
 
-The **AntiDeepfake** project provides a series of powerful self-supervised learning (SSL) models crafted to **detect deepfake speech** with state-of-the-art accuracy.
 
+The **AntiDeepfake** project provides a series of powerful foundation models post-trained for **deepfake detection**. The AntiDeepfake model can be used for feature extraction for deepfake detection in a zero-shot manner, or it may be further fine-tuned and optimized for a specific database or deepfake-related task.
 
 # 🤖 Available Models
 
@@ -191,13 +191,26 @@ Results shown below can be reproduced using scripts provided in our [GitHub repo
 | FakeOrReal          | 0.999   | 0.975    | 0.996     | 0.953  | 0.974    | 0.003 | 0.047 | 1.18 @ 0.1514        |
 | FakeOrReal-norm     | 0.999   | 0.979    | 0.966     | 0.993  | 0.979    | 0.033 | 0.007 | 1.73 @ 0.7434        |
 | In-the-Wild         | 0.999   | 0.985    | 0.993     | 0.983  | 0.988    | 0.011 | 0.017 | 1.31 @ 0.3748        |
-| WildSVDD-TEST_A     | 0.762   | 0.679    | 0.622     | 0.935  | 0.747    | 0.586 | 0.065 | 30.93 @ 0.9925       |
-| WildSVDD-TEST_B     | 0.718   | 0.714    | 0.754     | 0.846  | 0.797    | 0.546 | 0.154 | 36.11 @ 0.9838       |
+| Deepfake-Eval-2024  | 0.825   | 0.755    | 0.758     | 0.920  | 0.831    | 0.556 | 0.080 | 25.78 @ 0.9972       |
+
+
+
+You can also fine-tune this model on a specific database, the corresponding code is provided in our [GitHub repository](https://github.com/nii-yamagishilab/AntiDeepfake). Fine-tuning will follow a similar process to training a new model, except that model weights will be initialized as AntiDeepfake checkpoints.
+
+Below are the evaluation results of this model fine-tuned on the Deepfake-Eval-2024 training set and tested on its corresponding test set (as shown in the previous table):
+
+| Test Input Length | ROC AUC | Accuracy | Precision | Recall | F1-score | FPR    | FNR    | EER (%) @ Threshold |
+|-------------------|---------|----------|-----------|--------|----------|--------|--------|----------------------|
+| 4s                | 0.9211  | 0.7946   | 0.9486    | 0.7253 | 0.8221   | 0.0743 | 0.2747 | 19.82 @ 0.2562       |
+| 10s               | 0.9374  | 0.8298   | 0.9549    | 0.7756 | 0.8560   | 0.0687 | 0.2244 | 18.17 @ 0.2631       |
+| 13s               | 0.9401  | 0.8410   | 0.9574    | 0.7912 | 0.8664   | 0.0659 | 0.2088 | 18.19 @ 0.2653       |
+| 30s               | 0.9511  | 0.8678   | 0.9622    | 0.8281 | 0.8901   | 0.0595 | 0.1719 | 7.76  @ 0.2692       |
+| 50s               | 0.9541  | 0.8753   | 0.9543    | 0.8443 | 0.8959   | 0.0707 | 0.1557 | 8.91  @ 0.2393       |
 
 
 
 # **Training Set**
-Below is a breakdown of the training set used for fine-tuning.
+Below is a breakdown of the training set used for post-training of speech SSL models.
 
 | 📚 Database             | 🌍 Language         | ✅ Genuine (hrs) | ❌ Fake (hrs) |
 |----------------------|------------------|----------------|-------------|
@@ -242,5 +255,6 @@ This project is based on results obtained from project JPNP22007, commissioned b
 
 It is also partially supported by the following grants from the Japan Science and Technology Agency (JST):
 - AIP Acceleration Research (Grant No. JPMJCR24U3)
-- CREST (Grant No. JPMJCR20D3)
-- PRESTO (Grant No. JPMJPR23P9)
+- PRESTO (Grant No. JPMJPR23P9)
+
+This study was carried out using the TSUBAME4.0 supercomputer at Institute of Science Tokyo.