1 code implementation • Findings (ACL) 2022 • Haozhe An, Xiaojiang Liu, Donald Zhang
Pre-trained word embeddings, such as GloVe, have shown undesirable gender, racial, and religious biases.
no code implementations • 26 May 2023 • Haozhe An, Rachel Rudinger
We find that demographic attributes of a name (race, ethnicity, and gender) and name tokenization length are both factors that systematically affect the behavior of social commonsense reasoning models.
1 code implementation • 13 Oct 2022 • Haozhe An, Zongxia Li, Jieyu Zhao, Rachel Rudinger
A common limitation of diagnostic tests for detecting social biases in NLP models is that they may only detect stereotypic associations that are pre-specified by the designer of the test.
no code implementations • 25 May 2022 • Shramay Palta, Haozhe An, Yifan Yang, Shuaiyi Huang, Maharshi Gor
Retrieval based open-domain QA systems use retrieved documents and answer-span selection over retrieved documents to find best-answer candidates.
no code implementations • 6 Oct 2021 • Haoran Liu, Haoyi Xiong, Yaqing Wang, Haozhe An, Dongrui Wu, Dejing Dou
Specifically, we design a new metric $\mathcal{P}$-vector to represent the principal subspace of deep features learned in a DNN, and propose to measure angles between the principal subspaces using $\mathcal{P}$-vectors.
no code implementations • 1 Jan 2021 • Haozhe An, Haoyi Xiong, Xuhong LI, Xingjian Li, Dejing Dou, Zhanxing Zhu
The recent theoretical investigation (Li et al., 2020) on the upper bound of generalization error of deep neural networks (DNNs) demonstrates the potential of using the gradient norm as a measure that complements validation accuracy for model selection in practice.
no code implementations • 1 Jan 2021 • Haoran Liu, Haoyi Xiong, Yaqing Wang, Haozhe An, Dongrui Wu, Dejing Dou
While deep learning is effective to learn features/representations from data, the distributions of samples in feature spaces learned by various architectures for different training tasks (e. g., latent layers of AEs and feature vectors in CNN classifiers) have not been well-studied or compared.
no code implementations • 20 Jul 2020 • Xingjian Li, Haoyi Xiong, Haozhe An, Cheng-Zhong Xu, Dejing Dou
While the existing multitask learning algorithms need to run backpropagation over both the source and target datasets and usually consume a higher gradient complexity, XMixup transfers the knowledge from source to target tasks more efficiently: for every class of the target task, XMixup selects the auxiliary samples from the source dataset and augments training samples via the simple mixup strategy.
1 code implementation • ICML 2020 • Xingjian Li, Haoyi Xiong, Haozhe An, Cheng-Zhong Xu, Dejing Dou
RIFLE brings meaningful updates to the weights of deep CNN layers and improves low-level feature learning, while the effects of randomization can be easily converged throughout the overall learning procedure.
no code implementations • 26 Apr 2020 • Xingjian Li, Haoyi Xiong, Haozhe An, Dejing Dou, Chengzhong Xu
Softening labels of training datasets with respect to data representations has been frequently used to improve the training of deep neural networks (DNNs).