More Like this

1. On Separate Normalization in Self-supervised Transformers

   Chen, X; Wang, Y; Du, Y; Hassoun, S; Liu, L (December 2023, Curran Associates)

   Self-supervised training methods for transformers have demonstrated remarkable performance across various domains. Previous transformer-based models, such as masked autoencoders (MAE), typically utilize a single normalization layer for both the class token [CLS] and the tokens. We propose in this paper a new yet simple normalization method that separately normalizes embedding vectors respectively corresponding to normal tokens and the [CLS] token, in order to better capture their distinct characteristics and enhance downstream task performance. Our empirical study shows that the [CLS] embeddings learned with our separate normalization layer better encode the global contextual information and are distributed more uniformly in its anisotropic space. When the conventional normalization layer is replaced with a separate normalization layer, we observe an average 2.7% performance improvement in learning tasks from the image, natural language, and graph domains. 

   more » « less
2. Normalization and selecting non-differentially expressed genes improve machine learning modelling of cross-platform transcriptomic data

   Deng, Fei; Feng, Catherine H; Gao, Nan; Zhang, Lanjing (January 2025, arxiv.org)

   Normalization is a critical step in quantitative analyses of biological processes. Recent works show that cross-platform integration and normalization enable machine learning (ML) training on RNA microarray and RNA-seq data, but no independent datasets were used in their studies. Therefore, it is unclear how to improve ML modelling performance on independent RNA array and RNA-seq based datasets. Inspired by the house-keeping genes that are commonly used in experimental biology, this study tests the hypothesis that non-differentially expressed genes (NDEG) may improve normalization of transcriptomic data and subsequently cross-platform modelling performance of ML models. Microarray and RNA-seq datasets of the TCGA breast cancer were used as independent training and test datasets, respectively, to classify the molecular subtypes of breast cancer. NDEG (p>0.85) and differentially expressed genes (DEG, p<0.05) were selected based on the p values of ANOVA analysis and used for subsequent data normalization and classification, respectively. Models trained based on data from one platform were used for testing on the other platform. Our data show that NDEG and DEG gene selection could effectively improve the model classification performance. Normalization methods based on parametric statistical analysis were inferior to those based on nonparametric statistics. In this study, the LOG_QN and LOG_QNZ normalization methods combined with the neural network classification model seem to achieve better performance. Therefore, NDEG-based normalization appears useful for cross-platform testing on completely independent datasets. However, more studies are required to examine whether NDEG-based normalization can improve ML classification performance in other datasets and other omic data types. 

   more » « less
3. On Separate Normalization in Self-supervised Transformers

   Chen, Xiaohui; Wang, Yinkai; Du, Yuanqi; Hassoun, Soha; Liu, Li-Ping (December 2023, Advances in Neural Information Processing Systems 36)

   Self-supervised training methods for transformers have demonstrated remarkable performance across various domains. Previous transformer-based models, such as masked autoencoders (MAE), typically utilize a single normalization layer for both the [CLS] symbol and the tokens. We propose in this paper a simple modification that employs separate normalization layers for the tokens and the [CLS] symbol to better capture their distinct characteristics and enhance downstream task performance. Our method aims to alleviate the potential negative effects of using the same normalization statistics for both token types, which may not be optimally aligned with their individual roles. We empirically show that by utilizing a separate normalization layer, the [CLS] embeddings can better encode the global contextual information and are distributed more uniformly in its anisotropic space. When replacing the conventional normalization layer with the two separate layers, we observe an average 2.7% performance improvement over the image, natural language, and graph domains. 

   more » « less
4. Batch normalization followed by merging is powerful for phenotype prediction integrating multiple heterogeneous studies

   https://doi.org/10.1371/journal.pcbi.1010608  

   Gao, Yilin; Sun, Fengzhu (October 2023, PLOS Computational Biology)

   Roy, Sushmita (Ed.)

   Heterogeneity in different genomic studies compromises the performance of machine learning models in cross-study phenotype predictions. Overcoming heterogeneity when incorporating different studies in terms of phenotype prediction is a challenging and critical step for developing machine learning algorithms with reproducible prediction performance on independent datasets. We investigated the best approaches to integrate different studies of the same type of omics data under a variety of different heterogeneities. We developed a comprehensive workflow to simulate a variety of different types of heterogeneity and evaluate the performances of different integration methods together with batch normalization by using ComBat. We also demonstrated the results through realistic applications on six colorectal cancer (CRC) metagenomic studies and six tuberculosis (TB) gene expression studies, respectively. We showed that heterogeneity in different genomic studies can markedly negatively impact the machine learning classifier’s reproducibility. ComBat normalization improved the prediction performance of machine learning classifier when heterogeneous populations are present, and could successfully remove batch effects within the same population. We also showed that the machine learning classifier’s prediction accuracy can be markedly decreased as the underlying disease model became more different in training and test populations. Comparing different merging and integration methods, we found that merging and integration methods can outperform each other in different scenarios. In the realistic applications, we observed that the prediction accuracy improved when applying ComBat normalization with merging or integration methods in both CRC and TB studies. We illustrated that batch normalization is essential for mitigating both population differences of different studies and batch effects. We also showed that both merging strategy and integration methods can achieve good performances when combined with batch normalization. In addition, we explored the potential of boosting phenotype prediction performance by rank aggregation methods and showed that rank aggregation methods had similar performance as other ensemble learning approaches. 

   more » « less
5. Attentive Normalization

   https://doi.org/10.1007/978-3-030-58520-4_5  

   Li, X.; Sun, W.; Wu, T. (August 2020, 16th European Conference on Computer Vision (2020))

   Vedaldi, A. (Ed.)

   In state-of-the-art deep neural networks, both feature normalization and feature attention have become ubiquitous. They are usually studied as separate modules, however. In this paper, we propose a light-weight integration between the two schema and present Attentive Normalization (AN). Instead of learning a single affine transformation, AN learns a mixture of affine transformations and utilizes their weighted-sum as the final affine transformation applied to re-calibrate features in an instance-specific way. The weights are learned by leveraging channel-wise feature attention. In experiments, we test the proposed AN using four representative neural architectures. In the ImageNet-1000 classification benchmark and the MS-COCO 2017 object detection and instance segmentation benchmark. AN obtains consistent performance improvement for different neural architectures in both benchmarks with absolute increase of top-1 accuracy in ImageNet-1000 between 0.5\% and 2.7\%, and absolute increase up to 1.8\% and 2.2\% for bounding box and mask AP in MS-COCO respectively. We observe that the proposed AN provides a strong alternative to the widely used Squeeze-and-Excitation (SE) module. The source codes are publicly available at \href{https://github.com/iVMCL/AOGNet-v2}{the ImageNet Classification Repo} and \href{https://github.com/iVMCL/AttentiveNorm\_Detection}{the MS-COCO Detection and Segmentation Repo}. 

   more » « less