Recent progress of deep image classification models has provided great potential to improve state-of-the-art performance in related computer vision tasks. However, the transition to semantic segmentation is hampered by strict memory limitations of contemporary GPUs. The extent of feature map caching required by convolutional backprop poses significant challenges even for moderately sized Pascal images, while requiring careful architectural considerations when the source resolution is in the megapixel range. To address these concerns, we propose a novel DenseNet-based ladder-style architecture which features high modelling power and a very lean upsampling datapath. We also propose to substantially reduce the extent of feature map caching by exploiting inherent spatial efficiency of the DenseNet feature extractor. The resulting models deliver high performance with fewer parameters than competitive approaches, and allow training at megapixel resolution on commodity hardware. The presented experimental results outperform the state-of-the-art in terms of prediction accuracy and execution speed on Cityscapes, Pascal VOC 2012, CamVid and ROB 2018 datasets. Source code will be released upon publication.
Only learning one projection matrix from original samples to the corresponding binary labels is too strict and will consequentlly lose some intrinsic geometric structures of data. In this paper, we propose a novel transition subspace learning based least squares regression (TSL-LSR) model for multicategory image classification. The main idea of TSL-LSR is to learn a transition subspace between the original samples and binary labels to alleviate the problem of overfitting caused by strict projection learning. Moreover, in order to reflect the underlying low-rank structure of transition matrix and learn more discriminative projection matrix, a low-rank constraint is added to the transition subspace. Experimental results on several image datasets demonstrate the effectiveness of the proposed TSL-LSR model in comparison with state-of-the-art algorithms
In order to learn quickly with few samples, meta-learning utilizes prior knowledge learned from previous tasks. However, a critical challenge in meta-learning is task uncertainty and heterogeneity, which can not be handled via globally sharing knowledge among tasks. In this paper, based on gradient-based meta-learning, we propose a hierarchically structured meta-learning (HSML) algorithm that explicitly tailors the transferable knowledge to different clusters of tasks. Inspired by the way human beings organize knowledge, we resort to a hierarchical task clustering structure to cluster tasks. As a result, the proposed approach not only addresses the challenge via the knowledge customization to different clusters of tasks, but also preserves knowledge generalization among a cluster of similar tasks. To tackle the changing of task relationship, in addition, we extend the hierarchical structure to a continual learning environment. The experimental results show that our approach can achieve state-of-the-art performance in both toy-regression and few-shot image classification problems.
In most practical settings and theoretical analyses, one assumes that a model can be trained until convergence. However, the growing complexity of machine learning datasets and models may violate such assumptions. Indeed, current approaches for hyper-parameter tuning and neural architecture search tend to be limited by practical resource constraints. Therefore, we introduce a formal setting for studying training under the non-asymptotic, resource-constrained regime, i.e., budgeted training. We analyze the following problem: "given a dataset, algorithm, and fixed resource budget, what is the best achievable performance?" We focus on the number of optimization iterations as the representative resource. Under such a setting, we show that it is critical to adjust the learning rate schedule according to the given budget. Among budget-aware learning schedules, we find simple linear decay to be both robust and high-performing. We support our claim through extensive experiments with state-of-the-art models on ImageNet (image classification), Kinetics (video classification), MS COCO (object detection and instance segmentation), and Cityscapes (semantic segmentation). We also analyze our results and find that the key to a good schedule is budgeted convergence, a phenomenon whereby the gradient vanishes at the end of each allowed budget. We also revisit existing approaches for fast convergence and show that budget-aware learning schedules readily outperform such approaches under (the practical but under-explored) budgeted training setting.
In this paper, we address the hyperspectral image (HSI) classification task with a generative adversarial network and conditional random field (GAN-CRF) -based framework, which integrates a semi-supervised deep learning and a probabilistic graphical model, and make three contributions. First, we design four types of convolutional and transposed convolutional layers that consider the characteristics of HSIs to help with extracting discriminative features from limited numbers of labeled HSI samples. Second, we construct semi-supervised GANs to alleviate the shortage of training samples by adding labels to them and implicitly reconstructing real HSI data distribution through adversarial training. Third, we build dense conditional random fields (CRFs) on top of the random variables that are initialized to the softmax predictions of the trained GANs and are conditioned on HSIs to refine classification maps. This semi-supervised framework leverages the merits of discriminative and generative models through a game-theoretical approach. Moreover, even though we used very small numbers of labeled training HSI samples from the two most challenging and extensively studied datasets, the experimental results demonstrated that spectral-spatial GAN-CRF (SS-GAN-CRF) models achieved top-ranking accuracy for semi-supervised HSI classification.
In this letter, we propose a multitask deep learning method for classification of multiple hyperspectral data in a single training. Deep learning models have achieved promising results on hyperspectral image classification, but their performance highly rely on sufficient labeled samples, which are scarce on hyperspectral images. However, samples from multiple data sets might be sufficient to train one deep learning model, thereby improving its performance. To do so, we trained an identical feature extractor for all data, and the extracted features were fed into corresponding Softmax classifiers. Spectral knowledge was introduced to ensure that the shared features were similar across domains. Four hyperspectral data sets were used in the experiments. We achieved higher classification accuracies on three data sets (Pavia University, Pavia Center, and Indian Pines) and competitive results on the Salinas Valley data compared with the baseline. Spectral knowledge was useful to prevent the deep network from overfitting when the data shared similar spectral response. The proposed method tested on two deep CNNs successfully shows its ability to utilize samples from multiple data sets and enhance networks' performance.
In this paper, a novel statistical metric learning is developed for spectral-spatial classification of the hyperspectral image. First, the standard variance of the samples of each class in each batch is used to decrease the intra-class variance within each class. Then, the distances between the means of different classes are used to penalize the inter-class variance of the training samples. Finally, the standard variance between the means of different classes is added as an additional diversity term to repulse different classes from each other. Experiments have conducted over two real-world hyperspectral image datasets and the experimental results have shown the effectiveness of the proposed statistical metric learning.
Noisy labels are ubiquitous in real-world datasets, which poses a challenge for robustly training deep neural networks (DNNs) as DNNs usually have the high capacity to memorize the noisy labels. In this paper, we find that the test accuracy can be quantitatively characterized in terms of the noise ratio in datasets. In particular, the test accuracy is a quadratic function of the noise ratio in the case of symmetric noise, which explains the experimental findings previously published. Based on our analysis, we apply cross-validation to randomly split noisy datasets, which identifies most samples that have correct labels. Then we adopt the Co-teaching strategy which takes full advantage of the identified samples to train DNNs robustly against noisy labels. Compared with extensive state-of-the-art methods, our strategy consistently improves the generalization performance of DNNs under both synthetic and real-world training noise.
Regional dropout strategies have been proposed to enhance the performance of convolutional neural network classifiers. They have proved to be effective for guiding the model to attend on less discriminative parts of objects (e.g. leg as opposed to head of a person), thereby letting the network generalize better and have better object localization capabilities. On the other hand, current methods for regional dropout remove informative pixels on training images by overlaying a patch of either black pixels or random noise. Such removal is not desirable because it leads to information loss and inefficiency during training. We therefore propose the CutMix augmentation strategy: patches are cut and pasted among training images where the ground truth labels are also mixed proportionally to the area of the patches. By making efficient use of training pixels and retaining the regularization effect of regional dropout, CutMix consistently outperforms the state-of-the-art augmentation strategies on CIFAR and ImageNet classification tasks, as well as on the ImageNet weakly-supervised localization task. Moreover, unlike previous augmentation methods, our CutMix-trained ImageNet classifier, when used as a pretrained model, results in consistent performance gains in Pascal detection and MS-COCO image captioning benchmarks. We also show that CutMix improves the model robustness against input corruptions and its out-of-distribution detection performances. Source code and pretrained models are available at \href{https://github.com/clovaai/CutMix-PyTorch}{https://github.com/clovaai/CutMix-PyTorch}.
We investigate a classification problem using multiple mobile agents capable of collecting (partial) pose-dependent observations of an unknown environment. The objective is to classify an image over a finite time horizon. We propose a network architecture on how agents should form a local belief, take local actions, and extract relevant features from their raw partial observations. Agents are allowed to exchange information with their neighboring agents to update their own beliefs. It is shown how reinforcement learning techniques can be utilized to achieve decentralized implementation of the classification problem by running a decentralized consensus protocol. Our experimental results on the MNIST handwritten digit dataset demonstrates the effectiveness of our proposed framework.
Breast cancer is considered as the most fatal type of cancer among women worldwide and it is crucially important to be diagnosed at its early stages. In the current study, we aim to represent a fast and efficient framework which consists of two main parts:1- image classification, and 2- tumor region segmentation. At the initial stage, the images are classified into the two categories of normal and abnormal. Since the Deep Neural Networks have performed successfully in machine vision task, we would employ the convolutional neural networks for the classification of images. In the second stage, the suggested framework is to diagnose and segment the tumor in the mammography images. First, the mammography images are pre-processed by removing noise and artifacts, and then, segment the image using the level-set algorithm based on the spatial fuzzy c-means clustering. The proper initialization and optimal configuration have strong effects on the performance of the level-set segmentation. Thus, in our suggested framework, we have improved the level-set algorithm by utilizing the spatial fuzzy c-means clustering which ultimately results in a more precise segmentation. In order to evaluate the proposed approach, we conducted experiments using the Mammographic Image Analysis (MIAS) dataset. The tests have shown that the convolutional neural networks could achieve high accuracy in classification of images. Moreover, the improved level-set segmentation method, along with the fuzzy c-means clustering, could perfectly do the segmentation on the tumor area. The suggested method has classified the images with the accuracy of 78% and the AUC of 69%, which, as compared to the previous methods, is 2% more accurate and 6% better AUC; and has been able to extract the tumor area in a more precise way.
Can we automatically design a Convolutional Network (ConvNet) with the highest image classification accuracy under the latency constraint of a mobile device? Neural Architecture Search (NAS) for ConvNet design is a challenging problem due to the combinatorially large design space and search time (at least 200 GPU-hours). To alleviate this complexity, we propose Single-Path NAS, a novel differentiable NAS method for designing device-efficient ConvNets in less than 4 hours. 1. Novel NAS formulation: our method introduces a single-path, over-parameterized ConvNet to encode all architectural decisions with shared convolutional kernel parameters. 2. NAS efficiency: Our method decreases the NAS search cost down to 8 epochs (30 TPU-hours), i.e., up to 5,000x faster compared to prior work. 3. On-device image classification: Single-Path NAS achieves 74.96% top-1 accuracy on ImageNet with 79ms inference latency on a Pixel 1 phone, which is state-of-the-art accuracy compared to NAS methods with similar latency (<80ms).
A novel bilinear discriminant feature line analysis (BDFLA) is proposed for image feature extraction. The nearest feature line (NFL) is a powerful classifier. Some NFL-based subspace algorithms were introduced recently. In most of the classical NFL-based subspace learning approaches, the input samples are vectors. For image classification tasks, the image samples should be transformed to vectors first. This process induces a high computational complexity and may also lead to loss of the geometric feature of samples. The proposed BDFLA is a matrix-based algorithm. It aims to minimise the within-class scatter and maximise the between-class scatter based on a two-dimensional (2D) NFL. Experimental results on two-image databases confirm the effectiveness.
Convolutional networks for single-view object reconstruction have shown impressive performance and have become a popular subject of research. All existing techniques are united by the idea of having an encoder-decoder network that performs non-trivial reasoning about the 3D structure of the output space. In this work, we set up two alternative approaches that perform image classification and retrieval respectively. These simple baselines yield better results than state-of-the-art methods, both qualitatively and quantitatively. We show that encoder-decoder methods are statistically indistinguishable from these baselines, thus indicating that the current state of the art in single-view object reconstruction does not actually perform reconstruction but image classification. We identify aspects of popular experimental procedures that elicit this behavior and discuss ways to improve the current state of research.
Neural networks are known to be vulnerable to carefully crafted adversarial examples, and these malicious samples often transfer, i.e., they maintain their effectiveness even against other models. With great efforts delved into the transferability of adversarial examples, surprisingly, less attention has been paid to its impact on real-world deep learning deployment. In this paper, we investigate the transferability of adversarial examples across a wide range of real-world computer vision tasks, including image classification, explicit content detection, optical character recognition (OCR), and object detection. It represents the cybercriminal's situation where an ensemble of different detection mechanisms need to be evaded all at once. We propose practical attack that overcomes existing attacks' limitation of requiring task-specific loss functions by targeting on the `dispersion' of internal feature map. We report evaluation on four different computer vision tasks provided by Google Cloud Vision APIs to show how our approach outperforms existing attacks by degrading performance of multiple CV tasks by a large margin with only modest perturbations.
The performance of deep neural networks improves with more annotated data. The problem is that the budget for annotation is limited. One solution to this is active learning, where a model asks human to annotate data that it perceived as uncertain. A variety of recent methods have been proposed to apply active learning to deep networks but most of them are either designed specific for their target tasks or computationally inefficient for large networks. In this paper, we propose a novel active learning method that is simple but task-agnostic, and works efficiently with the deep networks. We attach a small parametric module, named "loss prediction module," to a target network, and learn it to predict target losses of unlabeled inputs. Then, this module can suggest data that the target model is likely to produce a wrong prediction. This method is task-agnostic as networks are learned from a single loss regardless of target tasks. We rigorously validate our method through image classification, object detection, and human pose estimation, with the recent network architectures. The results demonstrate that our method consistently outperforms the previous methods over the tasks.
In this paper, we are interested in boosting the representation capability of convolution neural networks which utilizing the inverted residual structure. Based on the success of Inverted Residual structure[Sandler et al. 2018] and Interleaved Low-Rank Group Convolutions[Sun et al. 2018], we rethink this two pattern of neural network structure, rather than NAS(Neural architecture search) method[Zoph and Le 2017; Pham et al. 2018; Liu et al. 2018b], we introduce uneven point-wise group convolution, which provide a novel search space for designing basic blocks to obtain better trade-off between representation capability and computational cost. Meanwhile, we propose two novel information flow patterns that will enable cross-group information flow for multiple group convolution layers with and without any channel permute/shuffle operation. Dense experiments on image classification task show that our proposed model, named Seesaw-Net, achieves state-of-the-art(SOTA) performance with limited computation and memory cost. Our code will be open-source and available together with pre-trained models.
This work tackles the problem of semi-supervised learning of image classifiers. Our main insight is that the field of semi-supervised learning can benefit from the quickly advancing field of self-supervised visual representation learning. Unifying these two approaches, we propose the framework of self-supervised semi-supervised learning and use it to derive two novel semi-supervised image classification methods. We demonstrate the effectiveness of these methods in comparison to both carefully tuned baselines, and existing semi-supervised learning methods. We then show that our approach and existing semi-supervised methods can be jointly trained, yielding a new state-of-the-art result on semi-supervised ILSVRC-2012 with 10% of labels.
Modern neural network training relies on piece-wise (sub-)differentiable functions in order to use backpropagation to update model parameters. In this work, we introduce a novel method to allow simple non-differentiable functions at intermediary layers of deep neural networks. We do so by training with a differentiable approximation bridge (DAB) neural network which approximates the non-differentiable forward function and provides gradient updates during backpropagation. We present strong empirical results (performing over 600 experiments) in four different domains: unsupervised (image) representation learning, variational (image) density estimation, image classification, and sequence sorting to demonstrate that our proposed method improves state of the art performance. We demonstrate that training with DAB aided discrete non-differentiable functions improves image reconstruction quality and posterior linear separability by 10% against the Gumbel-Softmax relaxed estimator [37, 26] as well as providing a 9% improvement in the test variational lower bound in comparison to the state of the art RELAX [16] discrete estimator. We also observe an accuracy improvement of 77% in neural sequence sorting and a 25% improvement against the straight-through estimator [5] in an image classification setting. The DAB network is not used for inference and expands the class of functions that are usable in neural networks.
In recent years, deep learning has presented a great advance in hyperspectral image (HSI) classification. Particularly, long short-term memory (LSTM), as a special deep learning structure, has shown great ability in modeling long-term dependencies in the time dimension of video or the spectral dimension of HSIs. However, the loss of spatial information makes it quite difficult to obtain the better performance. In order to address this problem, two novel deep models are proposed to extract more discriminative spatial-spectral features by exploiting the Convolutional LSTM (ConvLSTM). By taking the data patch in a local sliding window as the input of each memory cell band by band, the 2-D extended architecture of LSTM is considered for building the spatial-spectral ConvLSTM 2-D Neural Network (SSCL2DNN) to model long-range dependencies in the spectral domain. To better preserve the intrinsic structure information of the hyperspectral data, the spatial-spectral ConvLSTM 3-D Neural Network (SSCL3DNN) is proposed by extending LSTM to 3-D version for further improving the classification performance. The experiments, conducted on three commonly used HSI data sets, demonstrate that the proposed deep models have certain competitive advantages and can provide better classification performance than other state-of-the-art approaches.