diff --git a/README.md b/README.md
index 54283bf..d861284 100644
--- a/README.md
+++ b/README.md
@@ -2,7 +2,7 @@
 
 The `deepfdr` package provides a fully unsupervised, deep learning based FDR control method designed for spatial multiple testing. We utilize the [W-Net](https://arxiv.org/abs/1711.08506) and the concept of [local index of significance (LIS)](https://academic.oup.com/jrsssb/article/71/2/393/7092902) to connect unsupervised image segmentation to a 3D multiple testing problem that enables false discovery rate (FDR) control with a significant boost in statistical power. We tested our methodology using test statistics generated from FDG-PET data available at [ADNI database](https://adni.loni.usc.edu/). For specific details on methodology or training, please refer to our paper:
 
-T. Kim, H. Shu, Q. Jia, and M. de Leon (2023). [DeepFDR: A Deep Learning-based False Discovery Rate Control Method for Neuroimaging Data]( 	
+T. Kim, H. Shu, Q. Jia, and M. de Leon (2024). [DeepFDR: A Deep Learning-based False Discovery Rate Control Method for Neuroimaging Data]( 	
 https://proceedings.mlr.press/v238/kim24b.html). Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, PMLR 238:946-954, 2024.
 
 ## Table of Contents