PCIM: Learning pixel attributions via pixel-wise channel isolation mixing in high content imaging.

Deep neural networks have achieved great success in both computer vision and natural language processing tasks. How to improve the gradient flows is crucial in training very deep neural networks. To address this challenge, a gradient enhancement approach is proposed through constructing the short circuit neural connections. The proposed short circuit is a unidirectional neural connection that back propagates the sensitivities rather than gradients in neural networks from the deep layers to the shallow layers. Moreover, the short circuit is further formulated as a gradient truncation operation in its connecting layers, which can be plugged into the backbone models without introducing extra training parameters. Extensive experiments demonstrate that the deep neural networks, with the help of short circuit connection, gain a large margin of improvement over the baselines on both computer vision and natural language processing tasks. The work provides the promising solution to the low‐resource scenarios, such as, intelligence transport systems of computer vision, question answering of natural language processing.

Deep neural networks have attracted growing interests in medical image due to its success in computer vision tasks. One barrier for the application of deep neural networks is the need of large amounts of prior training pairs, which is not always feasible in clinical practice. Recently, the deep image prior framework shows that the convolutional neural network (CNN) can learn intrinsic structure information from the corrupted image. In this work, an iterative parametric reconstruction framework is proposed using deep neural network as constraint. The network does not need prior training pairs, but only the patient’s own CT image. The training is based on Logan plot derived from multi-bed-position dynamic positron emission tomography (PET) images using 68Ga-PRGD2 tracer. We formulated the estimation of the slope of Logan plot as a constraint optimization problem and solved it using the alternating direction method of multipliers (ADMM) algorithm. Quantification results based on real patient dataset shows that the proposed parametric reconstruction method is better than the Gaussian denoising and non-local mean denoising methods.

Recent advances in deep learning (DL) have led to great success in tasks of computer vision and pattern recognition. Sharing pre-trained DL models has been an important means to promote the rapid progress of research community and development of DL based systems. However, it also raises challenges to model authentication. It is quite necessary to protect the ownership of the DL models to be released. In this paper, we present a digital watermarking technique to deep neural networks (DNNs). We propose to mark a DNN by inserting an independent neural network that allows us to use selective weights for watermarking. The independent neural network is only used in the training phase and watermark verification phase, and will not be released publicly. Experiments have shown that, the performance of marked DNN on its original task will not be degraded significantly. Meantime, the watermark can be successfully embedded and extracted with a low neural network loss even under the common attacks including model fine-tuning and compression, which has shown the superiority and applicability of the proposed work.

Deep neural networks have accomplished enormous progress in tackling many problems. More specifically, convolutional neural network (CNN) is a category of deep networks that have been a dominant technique in computer vision tasks. Despite that these deep neural networks are highly effective; the ideal structure is still an issue that needs a lot of investigation. Deep Convolutional Neural Network model is usually designed manually by trials and repeated tests which enormously constrain its application. Many hyper-parameters of the CNN can affect the model performance. These parameters are depth of the network, numbers of convolutional layers, and numbers of kernels with their sizes. Therefore, it may be a huge challenge to design an appropriate CNN model that uses optimized hyper-parameters and reduces the reliance on manual involvement and domain expertise. In this paper, a design architecture method for CNNs is proposed by utilization of particle swarm optimization (PSO) algorithm to learn the optimal CNN hyper-parameters values. In the experiment, we used Modified National Institute of Standards and Technology (MNIST) database of handwritten digit recognition. The experiments showed that our proposed approach can find an architecture that is competitive to the state-of-the-art models with a testing error of 0.87%.

Fine-tuning deep convolutional neural networks for distinguishing illustrations from photographs

Transfer learning for deep neural networks is the process of first training a\nbase network on a source dataset, and then transferring the learned features\n(the network's weights) to a second network to be trained on a target dataset.\nThis idea has been shown to improve deep neural network's generalization\ncapabilities in many computer vision tasks such as image recognition and object\nlocalization. Apart from these applications, deep Convolutional Neural Networks\n(CNNs) have also recently gained popularity in the Time Series Classification\n(TSC) community. However, unlike for image recognition problems, transfer\nlearning techniques have not yet been investigated thoroughly for the TSC task.\nThis is surprising as the accuracy of deep learning models for TSC could\npotentially be improved if the model is fine-tuned from a pre-trained neural\nnetwork instead of training it from scratch. In this paper, we fill this gap by\ninvestigating how to transfer deep CNNs for the TSC task. To evaluate the\npotential of transfer learning, we performed extensive experiments using the\nUCR archive which is the largest publicly available TSC benchmark containing 85\ndatasets. For each dataset in the archive, we pre-trained a model and then\nfine-tuned it on the other datasets resulting in 7140 different deep neural\nnetworks. These experiments revealed that transfer learning can improve or\ndegrade the model's predictions depending on the dataset used for transfer.\nTherefore, in an effort to predict the best source dataset for a given target\ndataset, we propose a new method relying on Dynamic Time Warping to measure\ninter-datasets similarities. We describe how our method can guide the transfer\nto choose the best source dataset leading to an improvement in accuracy on 71\nout of 85 datasets.\n

Deep Neural Network (DNN) stands for multilayered Neural Network (NN) that is capable of progressively learn the more abstract and composite representations of the raw features of the input data received, with no need for any feature engineering. They are advanced NNs having repetitious hidden layers between the initial input and the final layer. The working principle of such a standard deep classifier is based on a hierarchy formed by the composition of linear functions and a defined nonlinear Activation Function (AF). It remains uncertain (not clear) how the DNN classifier can function so well. But it is clear from many studies that within DNN, the AF choice has a notable impact on the kinetics of training and the success of tasks. In the past few years, different AFs have been formulated. The choice of AF is still an area of active study. Hence, in this study, a novel deep Feed forward NN model with four AFs has been proposed for breast cancer classification: hidden layer 1: Swish, hidden layer, 2:-LeakyReLU, hidden layer 3: ReLU, and final output layer: naturally Sigmoidal. The purpose of the study is twofold. Firstly, this study is a step toward a more profound understanding of DNN with layer-wise different AFs. Secondly, research is also aimed to explore better DNN-based systems to build predictive models for breast cancer data with improved accuracy. Therefore, the benchmark UCI dataset WDBC was used for the validation of the framework and evaluated using a ten-fold CV method and various performance indicators. Multiple simulations and outcomes of the experimentations have shown that the proposed solution performs in a better way than the Sigmoid, ReLU, and LeakyReLU and Swish activation DNN in terms of different parameters. This analysis contributes to producing an expert and precise clinical dataset classification method for breast cancer. Furthermore, the model also achieved improved performance compared to many established state-of-the-art algorithms/models.

Although recent deep neural network algorithm has shown tremendous success in several computer vision tasks, their vulnerability against minute adversarial perturbations has raised a serious concern. In the early days of crafting these adversarial examples, artificial noises are optimized through the network and added in the images to decrease the confidence of the classifiers against the true class. However, recent efforts are showcasing the presence of natural adversarial examples which can also be effectively used to fool the deep neural networks with high confidence. In this paper, for the first time, we have raised the question that whether there is any robustness connection between artificial and natural adversarial examples. The possible robustness connection between natural and artificial adversarial examples is studied in the form that whether an adversarial example detector trained on artificial examples can detect the natural adversarial examples. We have analyzed several deep neural networks for the possible detection of artificial and natural adversarial examples in seen and unseen settings to set up a robust connection. The extensive experimental results reveal several interesting insights to defend the deep classifiers whether vulnerable against natural or artificially perturbed examples. We believe these findings can pave a way for the development of unified resiliency because defense against one attack is not sufficient for real-world use cases.

Recent advances in deep neural network have achieved remarkable successes in various computer vision tasks. However, deep neural network with millions of parameters may suffer from poor generalization due to overfitting. To improve the generalization performance, many methods have been proposed such as data augmentation, label smoothing and knowledge distillation. In this paper, we extend self-knowledge distillation to enhance the model generalization performance without incurring extra computation cost, called EWR-KD, which is a simple yet effective method to progressively distill knowledge from the model itself. Concretely, it consists of two components: 1) the self-distillation scheme that progressively softens the learning targets by using the past model prediction; 2) the sample-reweighting scheme that dynamically decides the trust degree to transfer more informative knowledge by introducing uncertainty estimation. With the two components, EWR-KD is robust to both corrupt noises and adversarial noises, and can be easily combined with current advanced regularization techniques. We theoretically show that EWR-KD minimizes cross-entropy by adding an epoch-wise regularization, which measures the difference between the past model prediction at ( <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$t-1$</tex> )-th epoch and the current prediction at <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$t$</tex> -th epoch. Finally, Extensive experimental results on clean datasets and noisy datasets empirically demonstrate that EWR-KD not only improves the performance of the state-of-the-art baseline but also yields well calibration.

Deep neural networks (DNNs) have achieved remarkable success in various computer vision tasks, e.g ., image classification. However, most of the existing models depend heavily on annotated data, where label noise is inevitable. Training with such noisy data negatively impacts the generalization performance of DNNs. To this end, recent advances in learning with noisy labels (LNL) adopt the sample selection strategy that identifies clean samples from the noisy dataset to update DNNs, using semi-supervised learning where rejected samples are treated as unlabeled data. However, existing LNL methods often overlook the varying fitting difficulties of different classes, resulting in suboptimal sample selection and confirmation bias, and consequently, the errors accumulate during semi-supervised training. In this paper, we propose a novel method, TNCollab, which aims at alleviating confirmation bias in both sample selection and semi-supervised training stages via two-network collaboration. Specifically, we introduce a class-adaptive threshold for sample selection to address the varying fitting difficulties across different classes. Additionally, we construct a hard set consisting of samples where the two networks disagree, and introduce a noise-robust loss to extract potentially useful information while maintaining robustness against label noise. Furthermore, we propose a dual consistency loss to ensure consistent predictions between the networks across different augmented views of the same sample, facilitating mutual learning. Extensive experiments demonstrate that TNCollab achieves state-of-the-art performance on image classification and facial expression recognition tasks, particularly on CIFAR-10, CIFAR-100, WebVision, Clothing1M, Tiny-ImageNet, and RAF-DB datasets, showing improved visual understanding and generalization capabilities. Our codes are available at https://github.com/Delete12137/TNCollab .

In a previous paper, we introduced a deep learning neural network that should be able to detect the existence of very shallow periodic planetary transits in the presence of red noise. The network in that feasibility study would not provide any further details about the detected transits. The current paper completes this missing part. We present a neural network that tags samples that were obtained during transits. This is essentially similar to the task of identifying the semantic context of each pixel in an image—an important task in computer vision, called “semantic segmentation,” which is often performed by deep neural networks. The neural network we present makes use of novel deep learning concepts such as U-Nets, Generative Adversarial Networks, and adversarial loss. The resulting segmentation should allow further studies of the light curves that are tagged as containing transits. This approach toward the detection and study of very shallow transits is bound to play a significant role in future space-based transit surveys such as PLATO, which are specifically aimed to detect those extremely difficult cases of long-period shallow transits. Our segmentation network also adds to the growing toolbox of deep learning approaches that are being increasingly used in the study of exoplanets; but, so far mainly for vetting transits, rather than their initial detection.

Deep learning is widely used for advanced applications of image and video processing with highperformance levels. Deep learning neural networks make use of the higher levels of accuracy inprediction and dynamic data analysis. Deep neural network has shown its extraordinaryperformance in different task of computer vision and machine learning tasks. These types ofnetworks often require large sets of labeled data for training and involve high computationalcomplexity. This poses considerable challenges for the development and deployment of deep neuralnetworks in realtime systems. In The proposed research work we analyzes the custom object-publicsector car and state government car detection and tracking system. Images frames from videosequence are used to detect moving vehicles based on Yolov3 object detection algorithm withdarknet frame work to trained own custom data model. And results display on webpage using FlaskWeb Framework. This deep learning method showed better classification and detecting ratecompare to background subtraction techniques. The percentage evolution of object detection rate isdiscussed in final result.

Object classification is a significant task in computer vision. It has become an effective research area as an important aspect of image processing and the building block of image localization, detection, and scene parsing. Object classification from low-quality images is difficult for the variance of object colors, aspect ratios, and cluttered backgrounds. The field of object classification has seen remarkable advancements, with the development of deep convolutional neural networks (DCNNs). Deep neural networks have been demonstrated as very powerful systems for facing the challenge of object classification from high-resolution images, but deploying such object classification networks on the embedded device remains challenging due to the high computational and memory requirements. Using high-quality images often causes high computational and memory complexity, whereas low-quality images can solve this issue. Hence, in this paper, we investigate an optimal architecture that accurately classifies low-quality images using DCNNs architectures. To validate different baselines on low-quality images, we perform experiments using webcam captured image datasets of 10 different objects. In this research work, we evaluate the proposed architecture by implementing popular CNN architectures. The experimental results validate that the MobileNet architecture delivers better than most of the available CNN architectures for low-resolution webcam image datasets.

&lt;p&gt;Deep Neural Networks (DNNs) have achieved state-of-the-art results in many computer vision tasks; however, DNNs have faced criticism for their lack of interpretability. Given the pervasiveness of DNNs in a multitude of applications, it is of paramount importance to fully understand the internal representations and behaviour of DNNs since safe and comprehensible utilization of DNN models is required before incorporating them into decision making processes for real-world applications. In this dissertation, we present several contributions towards understanding, interpreting, and learning representations in DNNs with an emphasis on studying absolute position information, interpreting latent representations to estimate certain semantic concepts, and learning robust representation. First, we study how much absolute position information is encoded in Convolutional Neural Networks (CNNs) as well as the source of this absolute position information. Our experiments reveal that a surprising degree of absolute position information is encoded in commonly used CNNs and zero padding enables CNNs to encode position information. Next, we analyze the relationship between boundary effects and padding in CNNs with respect to absolute position information. We also demonstrate how a CNN contains positional information in the latent representations if there exists a global pooling layer in the forward pass. We demonstrate that absolute position information is encoded based on the ordering of the channel dimensions, while semantic information is largely not. Second, we perform an empirical study on the ability of DNNs to encode shape information on a neuron-to-neuron and per-pixel level and show evidence that, while DNNs rely on texture information to recognize an object, a substantial amount of shape information is also encoded in DNNs. We further propose a new objective function for increasing a DNN’s ability to encode shape information by maximizing the mutual information between a network’s representations of two stylized images which share the same shape. Finally, we study the feature binding problem and present the first work which applies image blending to learn a robust representation for dense image labeling. Overall, we strongly believe the findings and demonstrated applications in this dissertation will benefit research areas concerned with understanding the different properties of DNNs.&lt;/p&gt;

&lt;p&gt;Deep Neural Networks (DNNs) have achieved state-of-the-art results in many computer vision tasks; however, DNNs have faced criticism for their lack of interpretability. Given the pervasiveness of DNNs in a multitude of applications, it is of paramount importance to fully understand the internal representations and behaviour of DNNs since safe and comprehensible utilization of DNN models is required before incorporating them into decision making processes for real-world applications. In this dissertation, we present several contributions towards understanding, interpreting, and learning representations in DNNs with an emphasis on studying absolute position information, interpreting latent representations to estimate certain semantic concepts, and learning robust representation. First, we study how much absolute position information is encoded in Convolutional Neural Networks (CNNs) as well as the source of this absolute position information. Our experiments reveal that a surprising degree of absolute position information is encoded in commonly used CNNs and zero padding enables CNNs to encode position information. Next, we analyze the relationship between boundary effects and padding in CNNs with respect to absolute position information. We also demonstrate how a CNN contains positional information in the latent representations if there exists a global pooling layer in the forward pass. We demonstrate that absolute position information is encoded based on the ordering of the channel dimensions, while semantic information is largely not. Second, we perform an empirical study on the ability of DNNs to encode shape information on a neuron-to-neuron and per-pixel level and show evidence that, while DNNs rely on texture information to recognize an object, a substantial amount of shape information is also encoded in DNNs. We further propose a new objective function for increasing a DNN’s ability to encode shape information by maximizing the mutual information between a network’s representations of two stylized images which share the same shape. Finally, we study the feature binding problem and present the first work which applies image blending to learn a robust representation for dense image labeling. Overall, we strongly believe the findings and demonstrated applications in this dissertation will benefit research areas concerned with understanding the different properties of DNNs.&lt;/p&gt;