Yi Li

Sepehr Dehdashtian, Ruozhen He, Yi Li et al.

Computer vision systems have witnessed rapid progress over the past two decades due to multiple advances in the field. As these systems are increasingly being deployed in high-stakes real-world applications, there is a dire need to ensure that they do not propagate or amplify any discriminatory tendencies in historical or human-curated data or inadvertently learn biases from spurious correlations. This paper presents a comprehensive survey on fairness that summarizes and sheds light on ongoing trends and successes in the context of computer vision. The topics we discuss include 1) The origin and technical definitions of fairness drawn from the wider fair machine learning literature and adjacent disciplines. 2) Work that sought to discover and analyze biases in computer vision systems. 3) A summary of methods proposed to mitigate bias in computer vision systems in recent years. 4) A comprehensive summary of resources and datasets produced by researchers to measure, analyze, and mitigate bias and enhance fairness. 5) Discussion of the field's success, continuing trends in the context of multimodal foundation and generative models, and gaps that still need to be addressed. The presented characterization should help researchers understand the importance of identifying and mitigating bias in computer vision and the state of the field and identify potential directions for future research.

Li Yi, Sheng Liu, Qi She et al.

Deep neural networks are able to memorize noisy labels easily with a softmax cross-entropy (CE) loss. Previous studies attempted to address this issue focus on incorporating a noise-robust loss function to the CE loss. However, the memorization issue is alleviated but still remains due to the non-robust CE loss. To address this issue, we focus on learning robust contrastive representations of data on which the classifier is hard to memorize the label noise under the CE loss. We propose a novel contrastive regularization function to learn such representations over noisy data where label noise does not dominate the representation learning. By theoretically investigating the representations induced by the proposed regularization function, we reveal that the learned representations keep information related to true labels and discard information related to corrupted labels. Moreover, our theoretical results also indicate that the learned representations are robust to the label noise. The effectiveness of this method is demonstrated with experiments on benchmark datasets.

Yi Li, Plamen Angelov, Neeraj Suri

Supervised learning-based adversarial attack detection methods rely on a large number of labeled data and suffer significant performance degradation when applying the trained model to new domains. In this paper, we propose a self-supervised representation learning framework for the adversarial attack detection task to address this drawback. Firstly, we map the pixels of augmented input images into an embedding space. Then, we employ the prototype-wise contrastive estimation loss to cluster prototypes as latent variables. Additionally, drawing inspiration from the concept of memory banks, we introduce a discrimination bank to distinguish and learn representations for each individual instance that shares the same or a similar prototype, establishing a connection between instances and their associated prototypes. We propose a parallel axial-attention (PAA)-based encoder to facilitate the training process by parallel training over height- and width-axis of attention maps. Experimental results show that, compared to various benchmark self-supervised vision learning models and supervised adversarial attack detection methods, the proposed model achieves state-of-the-art performance on the adversarial attack detection task across a wide range of images.

Lixu Wang, Bingqi Shang, Yi Li et al.

Vision Transformers (ViTs), extensively pre-trained on large-scale datasets, have become essential to foundation models, allowing excellent performance on diverse downstream tasks with minimal adaptation. Consequently, there is growing interest in adapting pre-trained ViTs across various fields, including privacy-sensitive domains where clients are often reluctant to share their data. Existing adaptation methods typically require direct data access, rendering them infeasible under these constraints. A straightforward solution may be sending the pre-trained ViT to clients for local adaptation, which poses issues of model intellectual property protection and incurs heavy client computation overhead. To address these issues, we propose a novel split adaptation (SA) method that enables effective downstream adaptation while protecting data and models. SA, inspired by split learning (SL), segments the pre-trained ViT into a frontend and a backend, with only the frontend shared with the client for data representation extraction. But unlike regular SL, SA replaces frontend parameters with low-bit quantized values, preventing direct exposure of the model. SA allows the client to add bi-level noise to the frontend and the extracted data representations, ensuring data protection. Accordingly, SA incorporates data-level and model-level out-of-distribution enhancements to mitigate noise injection's impact on adaptation performance. Our SA focuses on the challenging few-shot adaptation and adopts patch retrieval augmentation for overfitting alleviation. Extensive experiments on multiple datasets validate SA's superiority over state-of-the-art methods and demonstrate its defense against advanced data reconstruction attacks while preventing model leakage with minimal computation cost on the client side. The source codes can be found at https://github.com/conditionWang/Split_Adaptation.

Yi Li, Yunan Wu, Aggelos K. Katsaggelos

The advancement of The Laser Interferometer Gravitational-Wave Observatory (LIGO) has significantly enhanced the feasibility and reliability of gravitational wave detection. However, LIGO's high sensitivity makes it susceptible to transient noises known as glitches, which necessitate effective differentiation from real gravitational wave signals. Traditional approaches predominantly employ fully supervised or semi-supervised algorithms for the task of glitch classification and clustering. In the future task of identifying and classifying glitches across main and auxiliary channels, it is impractical to build a dataset with manually labeled ground-truth. In addition, the patterns of glitches can vary with time, generating new glitches without manual labels. In response to this challenge, we introduce the Cross-Temporal Spectrogram Autoencoder (CTSAE), a pioneering unsupervised method for the dimensionality reduction and clustering of gravitational wave glitches. CTSAE integrates a novel four-branch autoencoder with a hybrid of Convolutional Neural Networks (CNN) and Vision Transformers (ViT). To further extract features across multi-branches, we introduce a novel multi-branch fusion method using the CLS (Class) token. Our model, trained and evaluated on the GravitySpy O3 dataset on the main channel, demonstrates superior performance in clustering tasks when compared to state-of-the-art semi-supervised learning methods. To the best of our knowledge, CTSAE represents the first unsupervised approach tailored specifically for clustering LIGO data, marking a significant step forward in the field of gravitational wave research. The code of this paper is available at https://github.com/Zod-L/CTSAE

Yi Li, Wei Ping

Breast cancer diagnosis often requires accurate detection of metastasis in lymph nodes through Whole-slide Images (WSIs). Recent advances in deep convolutional neural networks (CNNs) have shown significant successes in medical image analysis and particularly in computational histopathology. Because of the outrageous large size of WSIs, most of the methods divide one slide into lots of small image patches and perform classification on each patch independently. However, neighboring patches often share spatial correlations, and ignoring these spatial correlations may result in inconsistent predictions. In this paper, we propose a neural conditional random field (NCRF) deep learning framework to detect cancer metastasis in WSIs. NCRF considers the spatial correlations between neighboring patches through a fully connected CRF which is directly incorporated on top of a CNN feature extractor. The whole deep network can be trained end-to-end with standard back-propagation algorithm with minor computational overhead from the CRF component. The CNN feature extractor can also benefit from considering spatial correlations via the CRF component. Compared to the baseline method without considering spatial correlations, we show that the proposed NCRF framework obtains probability maps of patch predictions with better visual quality. We also demonstrate that our method outperforms the baseline in cancer metastasis detection on the Camelyon16 dataset and achieves an average FROC score of 0.8096 on the test set. NCRF is open sourced at https://github.com/baidu-research/NCRF.

Leyi Zhao, Yi Li

Self-supervised pre-training models have been used successfully in several machine learning domains. However, only a tiny amount of work is related to music. In our work, we treat a spectrogram of music as a series of patches and design a self-supervised model that captures the features of these sequential patches: Patchifier, which makes good use of self-supervised learning methods from both NLP and CV domains. We do not use labeled data for the pre-training process, only a subset of the MTAT dataset containing 16k music clips. After pre-training, we apply the model to several downstream tasks. Our model achieves a considerably acceptable result compared to other audio representation models. Meanwhile, our work demonstrates that it makes sense to consider audio as a series of patch segments.

Li Yi

Generating lyrics and poems is one of the essential downstream tasks in the Natural Language Processing (NLP) field. Current methods have performed well in some lyrics generation scenarios but need further improvements in tasks requiring fine control. We propose a novel method for generating ancient Chinese lyrics (Song Ci), a type of ancient lyrics that involves precise control of song structure. The system is equipped with a phrase retriever and a phrase connector. Based on an input prompt, the phrase retriever picks phrases from a database to construct a phrase pool. The phrase connector then selects a series of phrases from the phrase pool that minimizes a multi-term loss function that considers rhyme, song structure, and fluency. Experimental results show that our method can generate high-quality ancient Chinese lyrics while performing well on topic and song structure control. We also expect our approach to be generalized to other lyrics-generating tasks.

Yi Li, Yang Sun, Syed Mohsen Naqvi

In self-supervised learning, it is challenging to reduce the gap between the enhancement performance on the estimated and target speech signals with existed pre-tasks. In this paper, we propose a multi-task pre-training method to improve the speech enhancement performance with self-supervised learning. Within the pre-training autoencoder (PAE), only a limited set of clean speech signals are required to learn their latent representations. Meanwhile, to solve the limitation of single pre-task, the proposed masking module exploits the dereverberated mask and estimated ratio mask to denoise the mixture as the second pre-task. Different from the PAE, where the target speech signals are estimated, the downstream task autoencoder (DAE) utilizes a large number of unlabeled and unseen reverberant mixtures to generate the estimated mixtures. The trained DAE is shared by the learned representations and masks. Experimental results on a benchmark dataset demonstrate that the proposed method outperforms the state-of-the-art approaches.

Yi Li, Yang Sun, Syed Mohsen Naqvi

Recently, self-supervised learning (SSL) techniques have been introduced to solve the monaural speech enhancement problem. Due to the lack of using clean phase information, the enhancement performance is limited in most SSL methods. Therefore, in this paper, we propose a phase-aware self-supervised learning based monaural speech enhancement method. The latent representations of both amplitude and phase are studied in two decoders of the foundation autoencoder (FAE) with only a limited set of clean speech signals independently. Then, the downstream autoencoder (DAE) learns a shared latent space between the clean speech and mixture representations with a large number of unseen mixtures. A complex-cycle-consistent (CCC) mechanism is proposed to minimize the reconstruction loss between the amplitude and phase domains. Besides, it is noticed that if the speech features are extracted as the multi-resolution spectra, the desired information distributed in spectra of different scales can be studied to further boost the performance. The NOISEX and DAPS corpora are used to generate mixtures with different interferences to evaluate the efficacy of the proposed method. It is highlighted that the clean speech and mixtures fed in FAE and DAE are not paired. Both ablation and comparison experimental results show that the proposed method clearly outperforms the state-of-the-art approaches.

Yi Li, Yang Sun, Syed Mohsen Naqvi

The state-of-the-art speech enhancement has limited performance in speech estimation accuracy. Recently, in deep learning, the Transformer shows the potential to exploit the long-range dependency in speech by self-attention. Therefore, it is introduced in speech enhancement to improve the speech estimation accuracy from a noise mixture. However, to address the computational cost issue in Transformer with self-attention, the axial attention is the option i.e., to split a 2D attention into two 1D attentions. Inspired by the axial attention, in the proposed method we calculate the attention map along both time- and frequency-axis to generate time and frequency sub-attention maps. Moreover, different from the axial attention, the proposed method provides two parallel multi-head attentions for time- and frequency-axis. Furthermore, it is proven in the literature that the lower frequency-band in speech, generally, contains more desired information than the higher frequency-band, in a noise mixture. Therefore, the frequency-band aware attention is proposed i.e., high frequency-band attention (HFA), and low frequency-band attention (LFA). The U-shaped Transformer is also first time introduced in the proposed method to further improve the speech estimation accuracy. The extensive evaluations over four public datasets, confirm the efficacy of the proposed method.

Yi Li, Yang Sun, Kirill Horoshenkov et al.

As a category of transfer learning, domain adaptation plays an important role in generalizing the model trained in one task and applying it to other similar tasks or settings. In speech enhancement, a well-trained acoustic model can be exploited to obtain the speech signal in the context of other languages, speakers, and environments. Recent domain adaptation research was developed more effectively with various neural networks and high-level abstract features. However, the related studies are more likely to transfer the well-trained model from a rich and more diverse domain to a limited and similar domain. Therefore, in this study, the domain adaptation method is proposed in unsupervised speech enhancement for the opposite circumstance that transferring to a larger and richer domain. On the one hand, the importance-weighting (IW) approach is exploited with a variance constrained autoencoder to reduce the shift of shared weights between the source and target domains. On the other hand, in order to train the classifier with the worst-case weights and minimize the risk, the minimax method is proposed. Both the proposed IW and minimax methods are evaluated from the VOICE BANK and IEEE datasets to the TIMIT dataset. The experiment results show that the proposed methods outperform the state-of-the-art approaches.

Xinyu Gao, Yi Li, Yanqing Qiu et al.

Multiple optical scattering occurs when light propagates in a non-uniform medium. During the multiple scattering, images were distorted and the spatial information they carried became scrambled. However, the image information is not lost but presents in the form of speckle patterns (SPs). In this study, we built up an optical random scattering system based on an LCD and an RGB laser source. We found that the image classification can be improved by the help of random scattering which is considered as a feedforward neural network to extracts features from image. Along with the ridge classification deployed on computer, we achieved excellent classification accuracy higher than 94%, for a variety of data sets covering medical, agricultural, environmental protection and other fields. In addition, the proposed optical scattering system has the advantages of high speed, low power consumption, and miniaturization, which is suitable for deploying in edge computing applications.

Kailong Liu, Xiaosong Hu, Zhongbao Wei et al.

This paper presents the development of machine learning-enabled data-driven models for effective capacity predictions for lithium-ion batteries under different cyclic conditions. To achieve this, a model structure is first proposed with the considerations of battery ageing tendency and the corresponding operational temperature and depth-of-discharge. Then based on a systematic understanding of covariance functions within the Gaussian process regression, two related data-driven models are developed. Specifically, by modifying the isotropic squared exponential kernel with an automatic relevance determination structure, 'Model A' could extract the highly relevant input features for capacity predictions. Through coupling the Arrhenius law and a polynomial equation into a compositional kernel, 'Model B' is capable of considering the electrochemical and empirical knowledge of battery degradation. The developed models are validated and compared on the Nickel Manganese Cobalt Oxide (NMC) lithium-ion batteries with various cycling patterns. Experimental results demonstrate that the modified Gaussian process regression model considering the battery electrochemical and empirical ageing signature outperforms other counterparts and is able to achieve satisfactory results for both one-step and multi-step predictions. The proposed technique is promising for battery capacity predictions under various cycling cases.

Hao Zhu, Huaibo Huang, Yi Li et al.

Talking face generation aims to synthesize a face video with precise lip synchronization as well as a smooth transition of facial motion over the entire video via the given speech clip and facial image. Most existing methods mainly focus on either disentangling the information in a single image or learning temporal information between frames. However, cross-modality coherence between audio and video information has not been well addressed during synthesis. In this paper, we propose a novel arbitrary talking face generation framework by discovering the audio-visual coherence via the proposed Asymmetric Mutual Information Estimator (AMIE). In addition, we propose a Dynamic Attention (DA) block by selectively focusing the lip area of the input image during the training stage, to further enhance lip synchronization. Experimental results on benchmark LRW dataset and GRID dataset transcend the state-of-the-art methods on prevalent metrics with robust high-resolution synthesizing on gender and pose variations.

Jiaji Huang, Yi Li, Wei Ping et al.

We propose a large margin criterion for training neural language models. Conventionally, neural language models are trained by minimizing perplexity (PPL) on grammatical sentences. However, we demonstrate that PPL may not be the best metric to optimize in some tasks, and further propose a large margin formulation. The proposed method aims to enlarge the margin between the "good" and "bad" sentences in a task-specific sense. It is trained end-to-end and can be widely applied to tasks that involve re-scoring of generated text. Compared with minimum-PPL training, our method gains up to 1.1 WER reduction for speech recognition and 1.0 BLEU increase for machine translation.

Zhiwei Lin, Yi Li, Xiaolian Guo

A ranking is an ordered sequence of items, in which an item with higher ranking score is more preferred than the items with lower ranking scores. In many information systems, rankings are widely used to represent the preferences over a set of items or candidates. The consensus measure of rankings is the problem of how to evaluate the degree to which the rankings agree. The consensus measure can be used to evaluate rankings in many information systems, as quite often there is not ground truth available for evaluation. This paper introduces a novel approach for consensus measure of rankings by using graph representation, in which the vertices or nodes are the items and the edges are the relationship of items in the rankings. Such representation leads to various algorithms for consensus measure in terms of different aspects of rankings, including the number of common patterns, the number of common patterns with fixed length and the length of the longest common patterns. The proposed measure can be adopted for various types of rankings, such as full rankings, partial rankings and rankings with ties. This paper demonstrates how the proposed approaches can be used to evaluate the quality of rank aggregation and the quality of top-$k$ rankings from Google and Bing search engines.