Neural network ensembles have been effectively used to improve generalization by combining the predictions of multiple independently trained models. However, the growing scale and complexity of deep neural networks have led to these methods becoming prohibitively expensive and time consuming to implement. Low-cost ensemble methods have become increasingly important as they can alleviate the need to train multiple models from scratch while retaining the generalization benefits that traditional ensemble learning methods afford. This dissertation introduces and formalizes a low-cost framework for constructing Subnetwork Ensembles, where a collection of child networks are formed by sampling, perturbing, and optimizing subnetworks from a trained parent model. We explore several distinct methodologies for generating child networks and we evaluate their efficacy through a variety of ablation studies and established benchmarks. Our findings reveal that this approach can greatly improve training efficiency, parametric utilization, and generalization performance while minimizing computational cost. Subnetwork Ensembles offer a compelling framework for exploring how we can build better systems by leveraging the unrealized potential of deep neural networks.
Monte Carlo (MC) dropout is a simple and efficient ensembling method that can improve the accuracy and confidence calibration of high-capacity deep neural network models. However, MC dropout is not as effective as more compute-intensive methods such as deep ensembles. This performance gap can be attributed to the relatively poor quality of individual models in the MC dropout ensemble and their lack of diversity. These issues can in turn be traced back to the coupled training and substantial parameter sharing of the dropout models. Motivated by this perspective, we propose a strategy to compute…
A biological neural network is constituted by numerous subnetworks and modules with different functionalities. For an artificial neural network, the relationship between a network and its subnetworks is also important and useful for both theoretical and algorithmic research, i.e. it can be exploited to develop incremental network training algorithm or parallel network training algorithm. In this paper we explore the relationship between an ELM neural network and its subnetworks. To the best of our knowledge, we are the first to prove a theorem that shows an ELM neural network can be scattered …
Ensemble Learning is an effective method for improving generalization in machine learning. However, as state-of-the-art neural networks grow larger, the computational cost associated with training several independent networks becomes expensive. We introduce a fast, low-cost method for creating diverse ensembles of neural networks without needing to train multiple models from scratch. We do this by first training a single parent network. We then create child networks by cloning the parent and dramatically pruning the parameters of each child to create an ensemble of members with unique and dive…
Brookhaven National Laboratory（布鲁克海文国家实验室）；Michigan State University（密歇根州立大学）；Texas A&M University（得克萨斯农工大学）；Oak Ridge National Laboratory（橡树岭国家实验室）；Argonne National Laboratory（阿贡国家实验室）
Ensembling is a popular method used to improve performance as a last resort. However, ensembling multiple models finetuned from a single pretrained model has been not very effective; this could be due to the lack of diversity among ensemble members. This paper proposes Multi-Ticket Ensemble, which finetunes different subnetworks of a single pretrained model and ensembles them. We empirically demonstrated that winning-ticket subnetworks produced more diverse predictions than dense networks, and their ensemble outperformed the standard ensemble on some tasks.
Beijing National Research Center for Information Science and Technology（北京信息科学与技术国家研究中心）；Tsinghua University（清华大学）
We propose a novel perspective to understand deep neural networks in an interpretable disentanglement form. For each semantic class, we extract a class-specific functional subnetwork from the original full model, with compressed structure while maintaining comparable prediction performance. The structure representations of extracted subnetworks display a resemblance to their corresponding class semantic similarities. We also apply extracted subnetworks in visual explanation and adversarial example detection tasks by merely replacing the original full model with class-specific subnetworks. Expe…
AI Squared, Inc.（AI Squared公司）；Johns Hopkins University（约翰斯·霍普金斯大学）
Neural networks have seen an explosion of usage and research in the past decade, particularly within the domains of computer vision and natural language processing. However, only recently have advancements in neural networks yielded performance improvements beyond narrow applications and translated to expanded multitask models capable of generalizing across multiple data types and modalities. Simultaneously, it has been shown that neural networks are overparameterized to a high degree, and pruning techniques have proved capable of significantly reducing the number of active weights within the …
Korea Advanced Institute of Science and Technology（韩国科学技术院）；Naver；AITRICS
Ensembles of deep neural networks have demonstrated superior performance, but their heavy computational cost hinders applying them for resource-limited environments. It motivates distilling knowledge from the ensemble teacher into a smaller student network, and there are two important design choices for this ensemble distillation: 1) how to construct the student network, and 2) what data should be shown during training. In this paper, we propose a weight averaging technique where a student with multiple subnetworks is trained to absorb the functional diversity of ensemble teachers, but then th…
We propose a general method for growing neural network with shared parameter by matching trained network to new input. By leveraging Hoeffding's inequality, we provide a theoretical base for improving performance by adding subnetwork to existing network. With the theoretical base of adding new subnetwork, we implement a matching method to apply trained subnetwork of existing network to new input. Our method has shown the ability to improve performance with higher parameter efficiency. It can also be applied to trans-task case and realize transfer learning by changing the combination of subnetw…
Gradient descent methods have long been the de facto standard for training deep neural networks. Millions of training samples are fed into models with billions of parameters, which are slowly updated over hundreds of epochs. Recently, it's been shown that large, randomly initialized neural networks contain subnetworks that perform as well as fully trained models. This insight offers a promising avenue for training future neural networks by simply pruning weights from large, random models. However, this problem is combinatorically hard and classical algorithms are not efficient at finding the b…
6th Generation (6G) industrial wireless subnetworks are expected to replace wired connectivity for control operation in robots and production modules. Interference management techniques such as centralized power control can improve spectral efficiency in dense deployments of such subnetworks. However, existing solutions for centralized power control may require full channel state information (CSI) of all the desired and interfering links, which may be cumbersome and time-consuming to obtain in dense deployments. This paper presents a novel solution for centralized power control for industrial …
Alternating minimization methods have recently been proposed as alternatives to the gradient descent for deep neural network optimization. Alternating minimization methods can typically decompose a deep neural network into layerwise subproblems, which can then be optimized in parallel. Despite the significant parallelism, alternating minimization methods are rarely explored in training deep neural networks because of the severe accuracy degradation. In this paper, we analyze the reason and propose to achieve a compelling trade-off between parallelism and accuracy by a reformulation called Tuna…
Allen Institute for Artificial Intelligence（艾伦人工智能研究所）；University of Washington（华盛顿大学）
Training a neural network is synonymous with learning the values of the weights. By contrast, we demonstrate that randomly weighted neural networks contain subnetworks which achieve impressive performance without ever training the weight values. Hidden in a randomly weighted Wide ResNet-50 we show that there is a subnetwork (with random weights) that is smaller than, but matches the performance of a ResNet-34 trained on ImageNet. Not only do these "untrained subnetworks" exist, but we provide an algorithm to effectively find them. We empirically show that as randomly weighted neural networks w…
NTT Computer and Data Science Laboratories, NTT Corporation（NTT公司NTT计算机与数据科学实验室）；Kyoto University（京都大学）
Few-shot learning for neural networks (NNs) is an important problem that aims to train NNs with a few data. The main challenge is how to avoid overfitting since over-parameterized NNs can easily overfit to such small dataset. Previous work (e.g. MAML by Finn et al. 2017) tackles this challenge by meta-learning, which learns how to learn from a few data by using various tasks. On the other hand, one conventional approach to avoid overfitting is restricting hypothesis spaces by endowing sparse NN structures like convolution layers in computer vision. However, although such manually-designed spar…
University of Texas Southwestern Medical Center（德克萨斯大学西南医学中心）
We present two novel hyperparameter optimization strategies for optimization of deep learning models with a modular architecture constructed of multiple subnetworks. As complex networks with multiple subnetworks become more frequently applied in machine learning, hyperparameter optimization methods are required to efficiently optimize their hyperparameters. Existing hyperparameter searches are general, and can be used to optimize such networks, however, by exploiting the multi-subnetwork architecture, these searches can be sped up substantially. The proposed methods offer faster convergence to…
Recent strategies achieved ensembling "for free" by fitting concurrently diverse subnetworks inside a single base network. The main idea during training is that each subnetwork learns to classify only one of the multiple inputs simultaneously provided. However, the question of how to best mix these multiple inputs has not been studied so far. In this paper, we introduce MixMo, a new generalized framework for learning multi-input multi-output deep subnetworks. Our key motivation is to replace the suboptimal summing operation hidden in previous approaches by a more appropriate mixing mechanism. …
the University of Texas, Austin（得克萨斯大学奥斯汀分校）；the University of Chicago（芝加哥大学）；Google Research（谷歌研究院）
Recent empirical works show that large deep neural networks are often highly redundant and one can find much smaller subnetworks without a significant drop of accuracy. However, most existing methods of network pruning are empirical and heuristic, leaving it open whether good subnetworks provably exist, how to find them efficiently, and if network pruning can be provably better than direct training using gradient descent. We answer these problems positively by proposing a simple greedy selection approach for finding good subnetworks, which starts from an empty network and greedily adds importa…
Once-for-All (OFA) is a Neural Architecture Search (NAS) framework designed to address the problem of searching efficient architectures for devices with different resources constraints by decoupling the training and the searching stages. The computationally expensive process of training the OFA neural network is done only once, and then it is possible to perform multiple searches for subnetworks extracted from this trained network according to each deployment scenario. In this work we aim to give one step further in the search for efficiency by explicitly conceiving the search stage as a multi…
Facebook AI Research（脸书人工智能研究院）；Univ. Grenoble Alpes（格勒诺布尔阿尔卑斯大学）；Inria（法国国家信息与自动化研究所）；CNRS（法国国家科学研究中心）；Grenoble INP（格勒诺布尔国立理工学院）
Convolutional neural networks trained without supervision come close to matching performance with supervised pre-training, but sometimes at the cost of an even higher number of parameters. Extracting subnetworks from these large unsupervised convnets with preserved performance is of particular interest to make them less computationally intensive. Typical pruning methods operate during training on a task while trying to maintain the performance of the pruned network on the same task. However, in self-supervised feature learning, the training objective is agnostic on the representation transfera…
The University of Texas at Austin（得克萨斯大学奥斯汀分校）；Facebook Assistant
The Lottery Ticket Hypothesis suggests large, over-parameterized neural networks consist of small, sparse subnetworks that can be trained in isolation to reach a similar (or better) test accuracy. However, the initialization and generalizability of the obtained sparse subnetworks have been recently called into question. Our work focuses on evaluating the initialization of sparse subnetworks under distributional shifts. Specifically, we investigate the extent to which a sparse subnetwork obtained in a source domain can be re-trained in isolation in a dissimilar, target domain. In addition, we e…
University of Cambridge（剑桥大学）；Google Research（谷歌研究院）；Stanford University（斯坦福大学）；Harvard University（哈佛大学）
Recent approaches to efficiently ensemble neural networks have shown that strong robustness and uncertainty performance can be achieved with a negligible gain in parameters over the original network. However, these methods still require multiple forward passes for prediction, leading to a significant computational cost. In this work, we show a surprising result: the benefits of using multiple predictions can be achieved `for free' under a single model's forward pass. In particular, we show that, using a multi-input multi-output (MIMO) configuration, one can utilize a single model's capacity to…
Eindhoven University of Technology（埃因霍温理工大学）；University of Twente（特温特大学）
Recent works on sparse neural network training (sparse training) have shown that a compelling trade-off between performance and efficiency can be achieved by training intrinsically sparse neural networks from scratch. Existing sparse training methods usually strive to find the best sparse subnetwork possible in one single run, without involving any expensive dense or pre-training steps. For instance, dynamic sparse training (DST), is capable of reaching a competitive performance of dense training by iteratively evolving the sparse topology during the course of training. In this paper, we argue…
The Multi-Prize Lottery Ticket Hypothesis posits that randomly initialized neural networks contain several subnetworks that achieve comparable accuracy to fully trained models of the same architecture. However, current methods require that the network is sufficiently overparameterized. In this work, we propose a modification to two state-of-the-art algorithms (Edge-Popup and Biprop) that finds high-accuracy subnetworks with no additional storage cost or scaling. The algorithm, Iterative Weight Recycling, identifies subsets of important weights within a randomly initialized network for intra-la…
Allen Institute for AI（艾伦人工智能研究所）；University of Washington（华盛顿大学）
The success of neural networks has driven a shift in focus from feature engineering to architecture engineering. However, successful networks today are constructed using a small and manually defined set of building blocks. Even in methods of neural architecture search (NAS) the network connectivity patterns are largely constrained. In this work we propose a method for discovering neural wirings. We relax the typical notion of layers and instead enable channels to form connections independent of each other. This allows for a much larger space of possible networks. The wiring of our network is n…
With the motive of training all the parameters of a neural network, we study why and when one can achieve this by iteratively creating, training, and combining randomly selected subnetworks. Such scenarios have either implicitly or explicitly emerged in the recent literature: see e.g., the Dropout family of regularization techniques, or some distributed ML training protocols that reduce communication/computation complexities, such as the Independent Subnet Training protocol. While these methods are studied empirically and utilized in practice, they often enjoy partial or no theoretical support…
We formalize the notion of a pseudo-ensemble, a (possibly infinite) collection of child models spawned from a parent model by perturbing it according to some noise process. E.g., dropout (Hinton et. al, 2012) in a deep neural network trains a pseudo-ensemble of child subnetworks generated by randomly masking nodes in the parent network. We present a novel regularizer based on making the behavior of a pseudo-ensemble robust with respect to the noise process generating it. In the fully-supervised setting, our regularizer matches the performance of dropout. But, unlike dropout, our regularizer na…
Spatiotemporal information processing is fundamental to brain functions. The present study investigates a canonic neural network model for spatiotemporal pattern recognition. Specifically, the model consists of two modules, a reservoir subnetwork and a decision-making subnetwork. The former projects complex spatiotemporal patterns into spatially separated neural representations, and the latter reads out these neural representations via integrating information over time; the two modules are combined together via supervised-learning using known examples. We elucidate the working mechanism of the…
Energy disaggregation, known in the literature as Non-Intrusive Load Monitoring (NILM), is the task of inferring the power demand of the individual appliances given the aggregate power demand recorded by a single smart meter which monitors multiple appliances. In this paper, we propose a deep neural network that combines a regression subnetwork with a classification subnetwork for solving the NILM problem. Specifically, we improve the generalization capability of the overall architecture by including an encoder-decoder with a tailored attention mechanism in the regression subnetwork. The atten…
Domain-Decomposed Randomized Neural Networks for Partial Differential Equations in Unbounded Domains
School of Mathematics and Statistics, Xi’an Jiaotong University（西安交通大学数学与统计学院）；State Key Laboratory of Multiphase Flow in Power Engineering, Xi’an Jiaotong University（西安交通大学动力工程多相流国家重点实验室）
City University of Hong Kong（香港城市大学）；Tsinghua University（清华大学）；Microsoft Research Asia（微软亚洲研究院）；Chinese Academy of Sciences（中国科学院）
One-shot weight sharing methods have recently drawn great attention in neural architecture search due to high efficiency and competitive performance. However, weight sharing across models has an inherent deficiency, i.e., insufficient training of subnetworks in hypernetworks. To alleviate this problem, we present a simple yet effective architecture distillation method. The central idea is that subnetworks can learn collaboratively and teach each other throughout the training process, aiming to boost the convergence of individual models. We introduce the concept of prioritized path, which refer…
Indian Institute of Information Technology Sri City（斯里城印度信息技术学院）；Wipro AI Labs（威普罗人工智能实验室）
Contending hate speech in social media is one of the most challenging social problems of our time. There are various types of anti-social behavior in social media. Foremost of them is aggressive behavior, which is causing many social issues such as affecting the social lives and mental health of social media users. In this paper, we propose an end-to-end ensemble-based architecture to automatically identify and classify aggressive tweets. Tweets are classified into three categories - Covertly Aggressive, Overtly Aggressive, and Non-Aggressive. The proposed architecture is an ensemble of smalle…
Northeastern University（东北大学）；Shenyang University of Technology（沈阳工业大学）；The Chinese University of Hong Kong（香港中文大学）；Second Affiliated Hospital of Dalian Medical University（大连医科大学附属第二医院）；Central Hospital Affiliated to Shenyang Medical College（沈阳医学院附属中心医院）
Accurate airway extraction from computed tomography (CT) images is a critical step for planning navigation bronchoscopy and quantitative assessment of airway-related chronic obstructive pulmonary disease (COPD). The existing methods are challenging to sufficiently segment the airway, especially the high-generation airway, with the constraint of the limited label and cannot meet the clinical use in COPD. We propose a novel two-stage 3D contextual transformer-based U-Net for airway segmentation using CT images. The method consists of two stages, performing initial and refined airway segmentation…
Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of inference without compromising accuracy. However, contemporary experience is that the sparse architectures produced by pruning are difficult to train from the start, which would similarly improve training performance. We find that a standard pruning technique naturally uncovers subnetworks whose initializations made them capable of training effectively. Based on these results, we articulate the "lottery ticket hypothesis:"…
Despite their great success, neural networks still remain as black-boxes due to the lack of interpretability. Here we propose a new analyzing method, namely the weight pathway analysis (WPA), to make them transparent. We consider weights in pathways that link neurons longitudinally from input neurons to output neurons, or simply weight pathways, as the basic units for understanding a neural network, and decompose a neural network into a series of subnetworks of such weight pathways. A visualization scheme of the subnetworks is presented that gives longitudinal perspectives of the network like …
The Lottery Ticket Hypothesis is a conjecture that every large neural network contains a subnetwork that, when trained in isolation, achieves comparable performance to the large network. An even stronger conjecture has been proven recently: Every sufficiently overparameterized network contains a subnetwork that, at random initialization, but without training, achieves comparable accuracy to the trained large network. This latter result, however, relies on a number of strong assumptions and guarantees a polynomial factor on the size of the large network compared to the target function. In this …
PruneSymNet: A Symbolic Neural Network and Pruning Algorithm for Symbolic Regression
Institute of Semiconductors, Chinese Academy of Sciences（中国科学院半导体研究所）；University of Chinese Academy of Sciences（中国科学院大学）；Beijing Key Laboratory of Semiconductor Neural Network Intelligent Sensing and Computing Technology（北京半导体神经网络智能感知与计算技术重点实验室）
Northeastern University（东北大学）；Santa Clara University（圣克拉拉大学）；Meta Research（Meta研究院）
Fully exploiting the learning capacity of neural networks requires overparameterized dense networks. On the other side, directly training sparse neural networks typically results in unsatisfactory performance. Lottery Ticket Hypothesis (LTH) provides a novel view to investigate sparse network training and maintain its capacity. Concretely, it claims there exist winning tickets from a randomly initialized network found by iterative magnitude pruning and preserving promising trainability (or we say being in trainable condition). In this work, we regard the winning ticket from LTH as the subnetwo…
MIT CSAIL（麻省理工学院计算机科学与人工智能实验室）；University of Cambridge（剑桥大学）；Element AI；University of Toronto（多伦多大学）；Vector Institute
Pruning is a well-established technique for removing unnecessary structure from neural networks after training to improve the performance of inference. Several recent results have explored the possibility of pruning at initialization time to provide similar benefits during training. In particular, the "lottery ticket hypothesis" conjectures that typical neural networks contain small subnetworks that can train to similar accuracy in a commensurate number of steps. The evidence for this claim is that a procedure based on iterative magnitude pruning (IMP) reliably finds such subnetworks retroacti…
