arXiv subjects

Haozhao Wang

Publications and source records attributed to Haozhao Wang.

68 records · Page 4Linked to original sources

arXiv2023

MGR: Multi-generator Based Rationalization

Rationalization is to employ a generator and a predictor to construct a self-explaining NLP model in which the generator selects a subset of human-intelligible pieces of the input text to the following predictor. However, rationalization suffers from two key challenges, i.e., spurious correlation and degeneration, where the predictor overfits the spurious or meaningless pieces solely selected by the not-yet well-trained generator and in turn deteriorates the generator. Although many studies have been proposed to address the two challenges, they are usually designed separately and do not take both of them into account. In this paper, we propose a simple yet effective method named MGR to simultaneously solve the two problems. The key idea of MGR is to employ multiple generators such that the occurrence stability of real pieces is improved and more meaningful pieces are delivered to the predictor. Empirically, we show that MGR improves the F1 score by up to 20.9% as compared to state-of-the-art methods. Codes are available at https://github.com/jugechengzi/Rationalization-MGR .

cs.LG↗

arXiv2023

Decoupled Rationalization with Asymmetric Learning Rates: A Flexible Lipschitz Restraint

A self-explaining rationalization model is generally constructed by a cooperative game where a generator selects the most human-intelligible pieces from the input text as rationales, followed by a predictor that makes predictions based on the selected rationales. However, such a cooperative game may incur the degeneration problem where the predictor overfits to the uninformative pieces generated by a not yet well-trained generator and in turn, leads the generator to converge to a sub-optimal model that tends to select senseless pieces. In this paper, we theoretically bridge degeneration with the predictor's Lipschitz continuity. Then, we empirically propose a simple but effective method named DR, which can naturally and flexibly restrain the Lipschitz constant of the predictor, to address the problem of degeneration. The main idea of DR is to decouple the generator and predictor to allocate them with asymmetric learning rates. A series of experiments conducted on two widely used benchmarks have verified the effectiveness of the proposed method. Codes: \href{https://github.com/jugechengzi/Rationalization-DR}{https://github.com/jugechengzi/Rationalization-DR}.

cs.LG↗

arXiv2023

Transform-Equivariant Consistency Learning for Temporal Sentence Grounding

This paper addresses the temporal sentence grounding (TSG). Although existing methods have made decent achievements in this task, they not only severely rely on abundant video-query paired data for training, but also easily fail into the dataset distribution bias. To alleviate these limitations, we introduce a novel Equivariant Consistency Regulation Learning (ECRL) framework to learn more discriminative query-related frame-wise representations for each video, in a self-supervised manner. Our motivation comes from that the temporal boundary of the query-guided activity should be consistently predicted under various video-level transformations. Concretely, we first design a series of spatio-temporal augmentations on both foreground and background video segments to generate a set of synthetic video samples. In particular, we devise a self-refine module to enhance the completeness and smoothness of the augmented video. Then, we present a novel self-supervised consistency loss (SSCL) applied on the original and augmented videos to capture their invariant query-related semantic by minimizing the KL-divergence between the sequence similarity of two videos and a prior Gaussian distribution of timestamp distance. At last, a shared grounding head is introduced to predict the transform-equivariant query-guided segment boundaries for both the original and augmented videos. Extensive experiments on three challenging datasets (ActivityNet, TACoS, and Charades-STA) demonstrate both effectiveness and efficiency of our proposed ECRL framework.

cs.CV↗

arXiv2023

DualMix: Unleashing the Potential of Data Augmentation for Online Class-Incremental Learning

Online Class-Incremental (OCI) learning has sparked new approaches to expand the previously trained model knowledge from sequentially arriving data streams with new classes. Unfortunately, OCI learning can suffer from catastrophic forgetting (CF) as the decision boundaries for old classes can become inaccurate when perturbated by new ones. Existing literature have applied the data augmentation (DA) to alleviate the model forgetting, while the role of DA in OCI has not been well understood so far. In this paper, we theoretically show that augmented samples with lower correlation to the original data are more effective in preventing forgetting. However, aggressive augmentation may also reduce the consistency between data and corresponding labels, which motivates us to exploit proper DA to boost the OCI performance and prevent the CF problem. We propose the Enhanced Mixup (EnMix) method that mixes the augmented samples and their labels simultaneously, which is shown to enhance the sample diversity while maintaining strong consistency with corresponding labels. Further, to solve the class imbalance problem, we design an Adaptive Mixup (AdpMix) method to calibrate the decision boundaries by mixing samples from both old and new classes and dynamically adjusting the label mixing ratio. Our approach is demonstrated to be effective on several benchmark datasets through extensive experiments, and it is shown to be compatible with other replay-based techniques.

cs.LG↗

arXiv2023

From Deterioration to Acceleration: A Calibration Approach to Rehabilitating Step Asynchronism in Federated Optimization

In the setting of federated optimization, where a global model is aggregated periodically, step asynchronism occurs when participants conduct model training by efficiently utilizing their computational resources. It is well acknowledged that step asynchronism leads to objective inconsistency under non-i.i.d. data, which degrades the model's accuracy. To address this issue, we propose a new algorithm FedaGrac, which calibrates the local direction to a predictive global orientation. Taking advantage of the estimated orientation, we guarantee that the aggregated model does not excessively deviate from the global optimum while fully utilizing the local updates of faster nodes. We theoretically prove that FedaGrac holds an improved order of convergence rate than the state-of-the-art approaches and eliminates the negative effect of step asynchronism. Empirical results show that our algorithm accelerates the training and enhances the final accuracy.

cs.DC↗

arXiv2022

FedTune: A Deep Dive into Efficient Federated Fine-Tuning with Pre-trained Transformers

Federated Learning (FL) is an emerging paradigm that enables distributed users to collaboratively and iteratively train machine learning models without sharing their private data. Motivated by the effectiveness and robustness of self-attention-based architectures, researchers are turning to using pre-trained Transformers (i.e., foundation models) instead of traditional convolutional neural networks in FL to leverage their excellent transfer learning capabilities. Despite recent progress, how pre-trained Transformer models play a role in FL remains obscure, that is, how to efficiently fine-tune these pre-trained models in FL and how FL users could benefit from this new paradigm. In this paper, we explore this issue and demonstrate that the fine-tuned Transformers achieve extraordinary performance on FL, and that the lightweight fine-tuning method facilitates a fast convergence rate and low communication costs. Concretely, we conduct a rigorous empirical study of three tuning methods (i.e., modifying the input, adding extra modules, and adjusting the backbone) using two types of pre-trained models (i.e., vision-language models and vision models) for FL. Our experiments show that 1) Fine-tuning the bias term of the backbone performs best when relying on a strong pre-trained model; 2) The vision-language model (e.g., CLIP) outperforms the pure vision model (e.g., ViT) and is more robust to the few-shot settings; 3) Compared to pure local training, FL with pre-trained models has a higher accuracy because it alleviates the problem of over-fitting. We will release our code and encourage further exploration of pre-trained Transformers and FL.

cs.LG↗

arXiv2022

PMR: Prototypical Modal Rebalance for Multimodal Learning

Multimodal learning (MML) aims to jointly exploit the common priors of different modalities to compensate for their inherent limitations. However, existing MML methods often optimize a uniform objective for different modalities, leading to the notorious "modality imbalance" problem and counterproductive MML performance. To address the problem, some existing methods modulate the learning pace based on the fused modality, which is dominated by the better modality and eventually results in a limited improvement on the worse modal. To better exploit the features of multimodal, we propose Prototypical Modality Rebalance (PMR) to perform stimulation on the particular slow-learning modality without interference from other modalities. Specifically, we introduce the prototypes that represent general features for each class, to build the non-parametric classifiers for uni-modal performance evaluation. Then, we try to accelerate the slow-learning modality by enhancing its clustering toward prototypes. Furthermore, to alleviate the suppression from the dominant modality, we introduce a prototype-based entropy regularization term during the early training stage to prevent premature convergence. Besides, our method only relies on the representations of each modality and without restrictions from model structures and fusion methods, making it with great application potential for various scenarios.

cs.LG↗

arXiv2022

FR: Folded Rationalization with a Unified Encoder

Conventional works generally employ a two-phase model in which a generator selects the most important pieces, followed by a predictor that makes predictions based on the selected pieces. However, such a two-phase model may incur the degeneration problem where the predictor overfits to the noise generated by a not yet well-trained generator and in turn, leads the generator to converge to a sub-optimal model that tends to select senseless pieces. To tackle this challenge, we propose Folded Rationalization (FR) that folds the two phases of the rationale model into one from the perspective of text semantic extraction. The key idea of FR is to employ a unified encoder between the generator and predictor, based on which FR can facilitate a better predictor by access to valuable information blocked by the generator in the traditional two-phase model and thus bring a better generator. Empirically, we show that FR improves the F1 score by up to 10.3% as compared to state-of-the-art methods.

cs.LG↗

arXiv2022

Sign Bit is Enough: A Learning Synchronization Framework for Multi-hop All-reduce with Ultimate Compression

Traditional one-bit compressed stochastic gradient descent can not be directly employed in multi-hop all-reduce, a widely adopted distributed training paradigm in network-intensive high-performance computing systems such as public clouds. According to our theoretical findings, due to the cascading compression, the training process has considerable deterioration on the convergence performance. To overcome this limitation, we implement a sign-bit compression-based learning synchronization framework, Marsit. It prevents cascading compression via an elaborate bit-wise operation for unbiased sign aggregation and its specific global compensation mechanism for mitigating compression deviation. The proposed framework retains the same theoretical convergence rate as non-compression mechanisms. Experimental results demonstrate that Marsit reduces up to 35% training time while preserving the same accuracy as training without compression.

cs.LG↗

arXiv2021

Parameterized Knowledge Transfer for Personalized Federated Learning

In recent years, personalized federated learning (pFL) has attracted increasing attention for its potential in dealing with statistical heterogeneity among clients. However, the state-of-the-art pFL methods rely on model parameters aggregation at the server side, which require all models to have the same structure and size, and thus limits the application for more heterogeneous scenarios. To deal with such model constraints, we exploit the potentials of heterogeneous model settings and propose a novel training framework to employ personalized models for different clients. Specifically, we formulate the aggregation procedure in original pFL into a personalized group knowledge transfer training algorithm, namely, KT-pFL, which enables each client to maintain a personalized soft prediction at the server side to guide the others' local training. KT-pFL updates the personalized soft prediction of each client by a linear combination of all local soft predictions using a knowledge coefficient matrix, which can adaptively reinforce the collaboration among clients who own similar data distribution. Furthermore, to quantify the contributions of each client to others' personalized training, the knowledge coefficient matrix is parameterized so that it can be trained simultaneously with the models. The knowledge coefficient matrix and the model parameters are alternatively updated in each round following the gradient descent way. Extensive experiments on various datasets (EMNIST, Fashion\_MNIST, CIFAR-10) are conducted under different settings (heterogeneous models and data distributions). It is demonstrated that the proposed framework is the first federated learning paradigm that realizes personalized model training via parameterized group knowledge transfer while achieving significant performance gain comparing with state-of-the-art algorithms.

cs.LG↗

arXiv2021

Gradient Scheduling with Global Momentum for Non-IID Data Distributed Asynchronous Training

Distributed asynchronous offline training has received widespread attention in recent years because of its high performance on large-scale data and complex models. As data are distributed from cloud-centric to edge nodes, a big challenge for distributed machine learning systems is how to handle native and natural non-independent and identically distributed (non-IID) data for training. Previous asynchronous training methods do not have a satisfying performance on non-IID data because it would result in that the training process fluctuates greatly which leads to an abnormal convergence. We propose a gradient scheduling algorithm with partly averaged gradients and global momentum (GSGM) for non-IID data distributed asynchronous training. Our key idea is to apply global momentum and local average to the biased gradient after scheduling, in order to make the training process steady. Experimental results show that for non-IID data training under the same experimental conditions, GSGM on popular optimization algorithms can achieve a 20% increase in training stability with a slight improvement in accuracy on Fashion-Mnist and CIFAR-10 datasets. Meanwhile, when expanding distributed scale on CIFAR-100 dataset that results in sparse data distribution, GSGM can perform a 37% improvement on training stability. Moreover, only GSGM can converge well when the number of computing nodes grows to 30, compared to the state-of-the-art distributed asynchronous algorithms. At the same time, GSGM is robust to different degrees of non-IID data.

cs.DC↗

arXiv2020

On the Convergence of Quantized Parallel Restarted SGD for Central Server Free Distributed Training

Communication is a crucial phase in the context of distributed training. Because parameter server (PS) frequently experiences network congestion, recent studies have found that training paradigms without a centralized server outperform the traditional server-based paradigms in terms of communication efficiency. However, with the increasing growth of model sizes, these server-free paradigms are also confronted with substantial communication overhead that seriously deteriorates the performance of distributed training. In this paper, we focus on communication efficiency of two serverless paradigms, i.e., Ring All-Reduce (RAR) and gossip, by proposing the Quantized Parallel Restarted Stochastic Gradient Descent (QPRSGD), an algorithm that allows multiple local SGD updates before a global synchronization, in synergy with the quantization to significantly reduce the communication overhead. We establish the bound of accumulative errors according to the synchronization mode and the network topology, which is essential to ensure the convergence property. Under both aggregation paradigms, the algorithm achieves the linear speedup property with respect to the number of local updates as well as the number of workers. Remarkably, the proposed algorithm achieves a convergence rate $O(1/\sqrt{NK^2M})$ under the gossip paradigm and outperforms all existing compression methods, where $N$ is the times of global synchronizations, and $K$ is the number of local updates, while $M$ is the number of nodes. An empirical study on various machine learning models demonstrates that the communication overhead is reduced by 90\%, and the convergence speed is boosted by up to 18.6 times, in a low bandwidth network, in comparison with Parallel SGD.

math.OC↗

arXiv2020

Intermittent Pulling with Local Compensation for Communication-Efficient Federated Learning

Federated Learning is a powerful machine learning paradigm to cooperatively train a global model with highly distributed data. A major bottleneck on the performance of distributed Stochastic Gradient Descent (SGD) algorithm for large-scale Federated Learning is the communication overhead on pushing local gradients and pulling global model. In this paper, to reduce the communication complexity of Federated Learning, a novel approach named Pulling Reduction with Local Compensation (PRLC) is proposed. Specifically, each training node intermittently pulls the global model from the server in SGD iterations, resulting in that it is sometimes unsynchronized with the server. In such a case, it will use its local update to compensate the gap between the local model and the global model. Our rigorous theoretical analysis of PRLC achieves two important findings. First, we prove that the convergence rate of PRLC preserves the same order as the classical synchronous SGD for both strongly-convex and non-convex cases with good scalability due to the linear speedup with respect to the number of training nodes. Second, we show that PRLC admits lower pulling frequency than the existing pulling reduction method without local compensation. We also conduct extensive experiments on various machine learning models to validate our theoretical results. Experimental results show that our approach achieves a significant pulling reduction over the state-of-the-art methods, e.g., PRLC requiring only half of the pulling operations of LAG.

cs.LG↗

arXiv2019

Heterogeneity-aware Gradient Coding for Straggler Tolerance

Gradient descent algorithms are widely used in machine learning. In order to deal with huge volume of data, we consider the implementation of gradient descent algorithms in a distributed computing setting where multiple workers compute the gradient over some partial data and the master node aggregates their results to obtain the gradient over the whole data. However, its performance can be severely affected by straggler workers. Recently, some coding-based approaches are introduced to mitigate the straggler problem, but they are efficient only when the workers are homogeneous, i.e., having the same computation capabilities. In this paper, we consider that the workers are heterogeneous which are common in modern distributed systems. We propose a novel heterogeneity-aware gradient coding scheme which can not only tolerate a predetermined number of stragglers but also fully utilize the computation capabilities of heterogeneous workers. We show that this scheme is optimal when the computation capabilities of workers are estimated accurately. A variant of this scheme is further proposed to improve the performance when the estimations of the computation capabilities are not so accurate. We conduct our schemes for gradient descent based image classification on QingCloud clusters. Evaluation results show that our schemes can reduce the whole computation time by up to $3\times$ compared with a state-of-the-art coding scheme.

cs.DC↗