The nonlocal-based blocks are designed for capturing long-range spatial-temporal dependencies in computer vision tasks. Although having shown excellent performances, they lack the mechanism to encode the rich, structured information among elements in an image. In this paper, to theoretically analyze the property of these nonlocal-based blocks, we provide a unified approach to interpreting them, where we view them as a graph filter generated on a fully-connected graph. When the graph filter is approximated by Chebyshev polynomials, a generalized formulation can be derived for explaining the existing nonlocal-based blocks ($\mathit{e.g.,}$ nonlocal block, nonlocal stage, double attention block). Furthermore, we propose an efficient and robust spectral nonlocal block, which can be flexibly inserted into deep neural networks to catch the long-range dependencies between spatial pixels or temporal frames. Experimental results demonstrate the clear-cut improvements and practical applicabilities of the spectral nonlocal block on image classification (Cifar-10/100, ImageNet), fine-grained image classification (CUB-200), action recognition (UCF-101), and person re-identification (ILID-SVID, Mars, Prid-2011) tasks.
The goal of meta-learning is to train a model on a variety of learning tasks, such that it can adapt to new problems within only a few iterations. Here we propose a principled information-theoretic model that optimally partitions the underlying problem space such that specialized expert decision-makers solve the resulting sub-problems. To drive this specialization we impose the same kind of information processing constraints both on the partitioning and the expert decision-makers. We argue that this specialization leads to efficient adaptation to new tasks. To demonstrate the generality of our approach we evaluate three meta-learning domains: image classification, regression, and reinforcement learning.
An increasing number of software applications incorporate runtime Deep Neural Networks (DNNs) to process sensor data and return inference results to humans. Effective deployment of DNNs in these interactive scenarios requires meeting latency and accuracy constraints while minimizing energy, a problem exacerbated by common system dynamics. Prior approaches handle dynamics through either (1) system-oblivious DNN adaptation, which adjusts DNN latency/accuracy tradeoffs, or (2) application-oblivious system adaptation, which adjusts resources to change latency/energy tradeoffs. In contrast, this paper improves on the state-of-the-art by coordinating application- and system-level adaptation. ALERT, our runtime scheduler, uses a probabilistic model to detect environmental volatility and then simultaneously select both a DNN and a system resource configuration to meet latency, accuracy, and energy constraints. We evaluate ALERT on CPU and GPU platforms for image and speech tasks in dynamic environments. ALERT's holistic approach achieves more than 13% energy reduction, and 27% error reduction over prior approaches that adapt solely at the application or system level. Furthermore, ALERT incurs only 3% more energy consumption and 2% higher DNN-inference error than an oracle scheme with perfect application and system knowledge.
Adaptive Precision Training: Quantify Back Propagation in Neural Networks with Fixed-point Numbers. Recent emerged quantization technique has been applied to inference of deep neural networks for fast and efficient execution. However, directly applying quantization in training can cause significant accuracy loss, thus remaining an open challenge.
Supervised classification methods often assume the train and test data distributions are the same and that all classes in the test set are present in the training set. However, deployed classifiers often require the ability to recognize inputs from outside the training set as unknowns. This problem has been studied under multiple paradigms including out-of-distribution detection and open set recognition. For convolutional neural networks, there have been two major approaches: 1) inference methods to separate knowns from unknowns and 2) feature space regularization strategies to improve model robustness to outlier inputs. There has been little effort to explore the relationship between the two approaches and directly compare performance on anything other than small-scale datasets that have at most 100 categories. Using ImageNet-1K and Places-434, we identify novel combinations of regularization and specialized inference methods that perform best across multiple outlier detection problems of increasing difficulty level. We found that input perturbation and temperature scaling yield the best performance on large scale datasets regardless of the feature space regularization strategy. Improving the feature space by regularizing against a background class can be helpful if an appropriate background class can be found, but this is impractical for large scale image classification datasets.
In this work, we exploit convolutional neural networks (CNNs) for the classification of very high resolution (VHR) polarimetric SAR (PolSAR) data. Due to the significant appearance of heterogeneous textures within these data, not only polarimetric features but also structural tensors are exploited to feed CNN models. For deep networks, we use the SegNet model for semantic segmentation, which corresponds to pixelwise classification in remote sensing. Our experiments on the airborne F-SAR data show that for VHR PolSAR images, SegNet could provide high accuracy for the classification task; and introducing structural tensors together with polarimetric features as inputs could help the network to focus more on geometrical information to significantly improve the classification performance.
Model-agnostic meta-learners aim to acquire meta-learned parameters from similar tasks to adapt to novel tasks from the same distribution with few gradient updates. With the flexibility in the choice of models, those frameworks demonstrate appealing performance on a variety of domains such as few-shot image classification and reinforcement learning. However, one important limitation of such frameworks is that they seek a common initialization shared across the entire task distribution, substantially limiting the diversity of the task distributions that they are able to learn from. In this paper, we augment MAML with the capability to identify the mode of tasks sampled from a multimodal task distribution and adapt quickly through gradient updates. Specifically, we propose a multimodal MAML (MMAML) framework, which is able to modulate its meta-learned prior parameters according to the identified mode, allowing more efficient fast adaptation. We evaluate the proposed model on a diverse set of few-shot learning tasks, including regression, image classification, and reinforcement learning. The results not only demonstrate the effectiveness of our model in modulating the meta-learned prior in response to the characteristics of tasks but also show that training on a multimodal distribution can produce an improvement over unimodal training.
Hyperspectral image (HSI) classification has become a hot topic in the field of remote sensing. In general, the complex characteristics of hyperspectral data make the accurate classification of such data challenging for traditional machine learning methods. In addition, hyperspectral imaging often deals with an inherently nonlinear relation between the captured spectral information and the corresponding materials. In recent years, deep learning has been recognized as a powerful feature-extraction tool to effectively address nonlinear problems and widely used in a number of image processing tasks. Motivated by those successful applications, deep learning has also been introduced to classify HSIs and demonstrated good performance. This survey paper presents a systematic review of deep learning-based HSI classification literatures and compares several strategies for this topic. Specifically, we first summarize the main challenges of HSI classification which cannot be effectively overcome by traditional machine learning methods, and also introduce the advantages of deep learning to handle these problems. Then, we build a framework which divides the corresponding works into spectral-feature networks, spatial-feature networks, and spectral-spatial-feature networks to systematically review the recent achievements in deep learning-based HSI classification. In addition, considering the fact that available training samples in the remote sensing field are usually very limited and training deep networks require a large number of samples, we include some strategies to improve classification performance, which can provide some guidelines for future studies on this topic. Finally, several representative deep learning-based classification methods are conducted on real HSIs in our experiments.
This research project studies the impact of convolutional neural networks (CNN) in image classification tasks. We explore different architectures and training configurations with the use of ReLUs, Nesterov's accelerated gradient, dropout and maxout networks. We work with the CIFAR-10 dataset as part of a Kaggle competition to identify objects in images. Initial results show that CNNs outperform our baseline by acting as invariant feature detectors. Comparisons between different preprocessing procedures show better results for global contrast normalization and ZCA whitening. ReLUs are much faster than tanh units and outperform sigmoids. We provide extensive details about our training hyperparameters, providing intuition for their selection that could help enhance learning in similar situations. We design 4 models of convolutional neural networks that explore characteristics such as depth, number of feature maps, size and overlap of kernels, pooling regions, and different subsampling techniques. Results favor models of moderate depth that use an extensive number of parameters in both convolutional and dense layers. Maxout networks are able to outperform rectifiers on some models but introduce too much noise as the complexity of the fully-connected layers increases. The final discussion explains our results and provides additional techniques that could improve performance.
Graph-based semi-supervised learning has been shown to be one of the most effective approaches for classification tasks from a wide range of domains, such as image classification and text classification, as they can exploit the connectivity patterns between labeled and unlabeled samples to improve learning performance. In this work, we advance this effective learning paradigm towards a scenario where labeled data are severely limited. More specifically, we address the problem of graph-based semi-supervised learning in the presence of severely limited labeled samples, and propose a new framework, called {\em Shoestring}, that improves the learning performance through semantic transfer from these very few labeled samples to large numbers of unlabeled samples. In particular, our framework learns a metric space in which classification can be performed by computing the similarity to centroid embedding of each class. {\em Shoestring} is trained in an end-to-end fashion to learn to leverage the semantic knowledge of limited labeled samples as well as their connectivity patterns with large numbers of unlabeled samples simultaneously. By combining {\em Shoestring} with graph convolutional networks, label propagation and their recent label-efficient variations (IGCN and GLP), we are able to achieve state-of-the-art node classification performance in the presence of very few labeled samples. In addition, we demonstrate the effectiveness of our framework on image classification tasks in the few-shot learning regime, with significant gains on miniImageNet ($2.57\%\sim3.59\%$) and tieredImageNet ($1.05\%\sim2.70\%$).
Convolutional Neural Networks (CNNs) have become indispensable for solving machine learning tasks in speech recognition, computer vision, and other areas that involve high-dimensional data. A CNN filters the input feature using a network containing spatial convolution operators with compactly supported stencils. In practice, the input data and the hidden features consist of a large number of channels, which in most CNNs are fully coupled by the convolution operators. This coupling leads to immense computational cost in the training and prediction phase. In this paper, we introduce LeanConvNets that are derived by sparsifying fully-coupled operators in existing CNNs. Our goal is to improve the efficiency of CNNs by reducing the number of weights, floating point operations and latency times, with minimal loss of accuracy. Our lean convolution operators involve tuning parameters that controls the trade-off between the network's accuracy and computational costs. These convolutions can be used in a wide range of existing networks, and we exemplify their use in residual networks (ResNets). Using a range of benchmark problems from image classification and semantic segmentation, we demonstrate that the resulting LeanConvNet's accuracy is close to state-of-the-art networks while being computationally less expensive. In our tests, the lean versions of ResNet in most cases outperform comparable reduced architectures such as MobileNets and ShuffleNets.
Compressing DNNs is important for the real-world applications operating on resource-constrained devices. However, we typically observe drastic performance deterioration when changing model size after training is completed. Therefore, retraining is required to resume the performance of the compressed models suitable for different devices. In this paper, we propose Decomposable-Net (the network decomposable in any size), which allows flexible changes to model size without retraining. We decompose weight matrices in the DNNs via singular value decomposition and adjust ranks according to the target model size. Unlike the existing low-rank compression methods that specialize the model to a fixed size, we propose a novel backpropagation scheme that jointly minimizes losses for both of full- and low-rank networks. This enables not only to maintain the performance of a full-rank network {\it without retraining} but also to improve low-rank networks in multiple sizes. Additionally, we introduce a simple criterion for rank selection that effectively suppresses approximation error. In experiments on the ImageNet classification task, Decomposable-Net yields superior accuracy in a wide range of model sizes. In particular, Decomposable-Net achieves the top-1 accuracy of $73.2\%$ with $0.27\times$MACs with ResNet-50, compared to Tucker decomposition ($67.4\% / 0.30\times$), Trained Rank Pruning ($70.6\% / 0.28\times$), and universally slimmable networks ($71.4\% / 0.26\times$).
Multitask learning, i.e. taking advantage of the relatedness of individual tasks in order to improve performance on all of them, is a core challenge in the field of machine learning. We focus on matrix regression tasks where the rank of the weight matrix is constrained to reduce sample complexity. We introduce the common mechanism regression (CMR) model which assumes a shared left low-rank component across all tasks, but allows an individual per-task right low-rank component. This dramatically reduces the number of samples needed for accurate estimation. The problem of jointly recovering the common and the local components has a non-convex bi-linear structure. We overcome this hurdle and provide a provably beneficial non-iterative spectral algorithm. Appealingly, the solution has favorable behavior as a function of the number of related tasks and the small number of samples available for each one. We demonstrate the efficacy of our approach for the challenging task of remote river discharge estimation across multiple river sites, where data for each task is naturally scarce. In this scenario sharing a low-rank component between the tasks translates to a shared spectral reflection of the water, which is a true underlying physical model. We also show the benefit of the approach on the markedly different setting of image classification where the common component can be interpreted as the shared convolution filters.
We investigate two questions in this paper: First, we ask to what extent "MPC friendly" models are already supported by major Machine Learning frameworks such as TensorFlow or PyTorch. Prior works provide protocols that only work on fixed-point integers and specialized activation functions, two aspects that are not supported by popular Machine Learning frameworks, and the need for these specialized model representations means that it is hard, and often impossible, to use e.g., TensorFlow to design, train and test models that later have to be evaluated securely. Second, we ask to what extent the functionality for evaluating Neural Networks already exists in general-purpose MPC frameworks. These frameworks have received more scrutiny, are better documented and supported on more platforms. Furthermore, they are typically flexible in terms of the threat model they support. In contrast, most secure evaluation protocols in the literature are targeted to a specific threat model and their implementations are only a "proof-of-concept", making it very hard for their adoption in practice. We answer both of the above questions in a positive way: We observe that the quantization techniques supported by both TensorFlow, PyTorch and MXNet can provide models in a representation that can be evaluated securely; and moreover, that this evaluation can be performed by a general purpose MPC framework. We perform extensive benchmarks to understand the exact trade-offs between different corruption models, network sizes and efficiency. These experiments provide an interesting insight into cost between active and passive security, as well as honest and dishonest majority. Our work shows then that the separating line between existing ML frameworks and existing MPC protocols may be narrower than implicitly suggested by previous works.
Decentralized optimization is emerging as a viable alternative for scalable distributed machine learning, but also introduces new challenges in terms of synchronization costs. To this end, several communication-reduction techniques, such as non-blocking communication, quantization, and local steps, have been explored in the decentralized setting. Due to the complexity of analyzing optimization in such a relaxed setting, this line of work often assumes \emph{global} communication rounds, which require additional synchronization. In this paper, we consider decentralized optimization in the simpler, but harder to analyze, \emph{asynchronous gossip} model, in which communication occurs in discrete, randomly chosen pairings among nodes. Perhaps surprisingly, we show that a variant of SGD called \emph{SwarmSGD} still converges in this setting, even if \emph{non-blocking communication}, \emph{quantization}, and \emph{local steps} are all applied \emph{in conjunction}, and even if the node data distributions and underlying graph topology are both \emph{heterogenous}. Our analysis is based on a new connection with multi-dimensional load-balancing processes. We implement this algorithm and deploy it in a super-computing environment, showing that it can outperform previous decentralized methods in terms of end-to-end training time, and that it can even rival carefully-tuned large-batch SGD for certain tasks.
Deep learning has become popular in recent years primarily due to the powerful computing device such as GPUs. However, deploying these deep models to end-user devices, smart phones, or embedded systems with limited resources is challenging. To reduce the computation and memory costs, we propose a novel lightweight deep learning module by low-rank pointwise residual (LPR) convolution, called LPRNet. Essentially, LPR aims at using low-rank approximation in pointwise convolution to further reduce the module size, while keeping depthwise convolutions as the residual module to rectify the LPR module. This is critical when the low-rankness undermines the convolution process. We embody our design by replacing modules of identical input-output dimension in MobileNet and ShuffleNetv2. Experiments on visual recognition tasks including image classification and face alignment on popular benchmarks show that our LPRNet achieves competitive performance but with significant reduction of Flops and memory cost compared to the state-of-the-art deep models focusing on model compression.
The effectiveness of Convolutional Neural Networks (CNNs) has been substantially attributed to their built-in property of translation equivariance. However, CNNs do not have embedded mechanisms to handle other types of transformations. In this work, we pay attention to scale changes, which regularly appear in various tasks due to the changing distances between the objects and the camera. First, we introduce the general theory for building scale-equivariant convolutional networks with steerable filters. We develop scale-convolution and generalize other common blocks to be scale-equivariant. We demonstrate the computational efficiency and numerical stability of the proposed method. We compare the proposed models to the previously developed methods for scale equivariance and local scale invariance. We demonstrate state-of-the-art results on MNIST-scale dataset and on STL-10 dataset in the supervised learning setting.
In the last couple of years, several adversarial attack methods based on different threat models have been proposed for the image classification problem. Most existing defenses consider additive threat models in which sample perturbations have bounded L_p norms. These defenses, however, can be vulnerable against adversarial attacks under non-additive threat models. An example of an attack method based on a non-additive threat model is the Wasserstein adversarial attack proposed by Wong et al. (2019), where the distance between an image and its adversarial example is determined by the Wasserstein metric ("earth-mover distance") between their normalized pixel intensities. Until now, there has been no certifiable defense against this type of attack. In this work, we propose the first defense with certified robustness against Wasserstein Adversarial attacks using randomized smoothing. We develop this certificate by considering the space of possible flows between images, and representing this space such that Wasserstein distance between images is upper-bounded by L_1 distance in this flow-space. We can then apply existing randomized smoothing certificates for the L_1 metric. In MNIST and CIFAR-10 datasets, we find that our proposed defense is also practically effective, demonstrating significantly improved accuracy under Wasserstein adversarial attack compared to unprotected models.
Human beings are able to master a variety of knowledge and skills with ongoing learning. By contrast, dramatic performance degradation is observed when new tasks are added to an existing neural network model. This phenomenon, termed as \emph{Catastrophic Forgetting}, is one of the major roadblocks that prevent deep neural networks from achieving human-level artificial intelligence. Several research efforts, e.g. \emph{Lifelong} or \emph{Continual} learning algorithms, have been proposed to tackle this problem. However, they either suffer from an accumulating drop in performance as the task sequence grows longer, or require to store an excessive amount of model parameters for historical memory, or cannot obtain competitive performance on the new tasks. In this paper, we focus on the incremental multi-task image classification scenario. Inspired by the learning process of human students, where they usually decompose complex tasks into easier goals, we propose an adversarial feature alignment method to avoid catastrophic forgetting. In our design, both the low-level visual features and high-level semantic features serve as soft targets and guide the training process in multiple stages, which provide sufficient supervised information of the old tasks and help to reduce forgetting. Due to the knowledge distillation and regularization phenomenons, the proposed method gains even better performance than finetuning on the new tasks, which makes it stand out from other methods. Extensive experiments in several typical lifelong learning scenarios demonstrate that our method outperforms the state-of-the-art methods in both accuracies on new tasks and performance preservation on old tasks.
Deep neural networks achieve superior performance in challenging tasks such as image classification. However, deep classifiers tend to incorrectly classify out-of-distribution (OOD) inputs, which are inputs that do not belong to the classifier training distribution. Several approaches have been proposed to detect OOD inputs, but the detection task is still an ongoing challenge. In this paper, we propose a new OOD detection approach that can be easily applied to an existing classifier and does not need to have access to OOD samples. The detector is a one-class classifier trained on the output of an early layer of the original classifier fed with its original training set. We apply our approach to several low- and high-dimensional datasets and compare it to the state-of-the-art detection approaches. Our approach achieves substantially better results over multiple metrics.