اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدScaling Vision with Sparse Mixture of Experts
416
0
0.0
(
0
)
نشر من قبل
Carlos Riquelme Ruiz
تاريخ النشر
2021
مجال البحث
الهندسة المعلوماتية
والبحث باللغة
English
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Sparsely-gated Mixture of Experts networks (MoEs) have demonstrated excellent scalability in Natural Language Processing. In Computer Vision, however, almost all performant networks are dense, that is, every input is processed by every parameter. We present a Vision MoE (V-MoE), a sparse version of the Vision Transformer, that is scalable and competitive with the largest dense networks. When applied to image recognition, V-MoE matches the performance of state-of-the-art networks, while requiring as little as half of the compute at inference time. Further, we propose an extension to the routing algorithm that can prioritize subsets of each input across the entire batch, leading to adaptive per-image compute. This allows V-MoE to trade-off performance and compute smoothly at test-time. Finally, we demonstrate the potential of V-MoE to scale vision models, and train a 15B parameter model that attains 90.35% on ImageNet.
قيم البحث
اقرأ أيضاً
Mixture of Experts (MoE) is a popular framework for modeling heterogeneity in data for regression, classification, and clustering. For regression and cluster analyses of continuous data, MoE usually use normal experts following the Gaussian distribut
ion. However, for a set of data containing a group or groups of observations with heavy tails or atypical observations, the use of normal experts is unsuitable and can unduly affect the fit of the MoE model. We introduce a robust MoE modeling using the $t$ distribution. The proposed $t$ MoE (TMoE) deals with these issues regarding heavy-tailed and noisy data. We develop a dedicated expectation-maximization (EM) algorithm to estimate the parameters of the proposed model by monotonically maximizing the observed data log-likelihood. We describe how the presented model can be used in prediction and in model-based clustering of regression data. The proposed model is validated on numerical experiments carried out on simulated data, which show the effectiveness and the robustness of the proposed model in terms of modeling non-linear regression functions as well as in model-based clustering. Then, it is applied to the real-world data of tone perception for musical data analysis, and the one of temperature anomalies for the analysis of climate change data. The obtained results show the usefulness of the TMoE model for practical applications.
Mixture of Experts (MoE) is a popular framework in the fields of statistics and machine learning for modeling heterogeneity in data for regression, classification and clustering. MoE for continuous data are usually based on the normal distribution. H
owever, it is known that for data with asymmetric behavior, heavy tails and atypical observations, the use of the normal distribution is unsuitable. We introduce a new robust non-normal mixture of experts modeling using the skew $t$ distribution. The proposed skew $t$ mixture of experts, named STMoE, handles these issues of the normal mixtures experts regarding possibly skewed, heavy-tailed and noisy data. We develop a dedicated expectation conditional maximization (ECM) algorithm to estimate the model parameters by monotonically maximizing the observed data log-likelihood. We describe how the presented model can be used in prediction and in model-based clustering of regression data. Numerical experiments carried out on simulated data show the effectiveness and the robustness of the proposed model in fitting non-linear regression functions as well as in model-based clustering. Then, the proposed model is applied to the real-world data of tone perception for musical data analysis, and the one of temperature anomalies for the analysis of climate change data. The obtained results confirm the usefulness of the model for practical data analysis applications.
Many generation tasks follow a one-to-many mapping relationship: each input could be associated with multiple outputs. Existing methods like Conditional Variational AutoEncoder(CVAE) employ a latent variable to model this one-to-many relationship. Ho
wever, this high-dimensional and dense latent variable lacks explainability and usually leads to poor and uncontrollable generations. In this paper, we innovatively introduce the linguistic concept of pattern to decompose the one-to-many mapping into multiple one-to-one mappings and further propose a model named Sparse Pattern Mixture of Experts(SPMoE). Each one-to-one mapping is associated with a conditional generation pattern and is modeled with an expert in SPMoE. To ensure each language pattern can be exclusively handled with an expert model for better explainability and diversity, a sparse mechanism is employed to coordinate all the expert models in SPMoE. We assess the performance of our SPMoE on the paraphrase generation task and the experiment results prove that SPMoE can achieve a good balance in terms of quality, pattern-level diversity, and corpus-level diversity.
Domain generalizable (DG) person re-identification (ReID) is a challenging problem because we cannot access any unseen target domain data during training. Almost all the existing DG ReID methods follow the same pipeline where they use a hybrid datase
t from multiple source domains for training, and then directly apply the trained model to the unseen target domains for testing. These methods often neglect individual source domains discriminative characteristics and their relevances w.r.t. the unseen target domains, though both of which can be leveraged to help the models generalization. To handle the above two issues, we propose a novel method called the relevance-aware mixture of experts (RaMoE), using an effective voting-based mixture mechanism to dynamically leverage source domains diverse characteristics to improve the models generalization. Specifically, we propose a decorrelation loss to make the source domain networks (experts) keep the diversity and discriminability of individual domains characteristics. Besides, we design a voting network to adaptively integrate all the experts features into the more generalizable aggregated features with domain relevance. Considering the target domains invisibility during training, we propose a novel learning-to-learn algorithm combined with our relation alignment loss to update the voting network. Extensive experiments demonstrate that our proposed RaMoE outperforms the state-of-the-art methods.
Recently, Mixture of Experts (MoE) based Transformer has shown promising results in many domains. This is largely due to the following advantages of this architecture: firstly, MoE based Transformer can increase model capacity without computational c
ost increasing both at training and inference time. Besides, MoE based Transformer is a dynamic network which can adapt to the varying complexity of input instances in realworld applications. In this work, we explore the MoE based model for speech recognition, named SpeechMoE. To further control the sparsity of router activation and improve the diversity of gate values, we propose a sparsity L1 loss and a mean importance loss respectively. In addition, a new router architecture is used in SpeechMoE which can simultaneously utilize the information from a shared embedding network and the hierarchical representation of different MoE layers. Experimental results show that SpeechMoE can achieve lower character error rate (CER) with comparable computation cost than traditional static networks, providing 7.0%-23.0% relative CER improvements on four evaluation datasets.
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
