Deep neural networks are representation learning techniques. During training, a deep net is capable of generating a descriptive language of unprecedented size and detail in machine learning. Extracting the descriptive language coded within a trained CNN model (in the case of image data), and reusing it for other purposes is a field of interest, as it provides access to the visual descriptors previously learnt by the CNN after processing millions of images, without requiring an expensive training phase. Contributions to this field (commonly known as feature representation transfer or transfer learning) have been purely empirical so far, extracting all CNN features from a single layer close to the output and testing their performance by feeding them to a classifier. This approach has provided consistent results, although its relevance is limited to classification tasks. In a completely different approach, in this paper we statistically measure the discriminative power of every single feature found within a deep CNN, when used for characterizing every class of 11 datasets. We seek to provide new insights into the behavior of CNN features, particularly the ones from convolutional layers, as this can be relevant for their application to knowledge representation and reasoning. Our results confirm that low and middle level features may behave differently to high level features, but only under certain conditions. We find that all CNN features can be used for knowledge representation purposes both by their presence or by their absence, doubling the information a single CNN feature may provide. We also study how much noise these features may include, and propose a thresholding approach to discard most of it. All these insights have a direct application to the generation of CNN embedding spaces.

Cilia are hairlike structures protruding from nearly every cell in the body. Diseases known as ciliopathies, where cilia function is disrupted, can result in a wide spectrum of disorders. However, most techniques for assessing ciliary motion rely on manual identification and tracking of cilia; this process is laborious and error-prone, and does not scale well. Even where automated ciliary motion analysis tools exist, their applicability is limited. Here, we propose an end-to-end computational machine learning pipeline that automatically identifies regions of cilia from videos, extracts patches of cilia, and classifies patients as exhibiting normal or abnormal ciliary motion. In particular, we demonstrate how convolutional LSTM are able to encode complex features while remaining sensitive enough to differentiate between a variety of motion patterns. Our framework achieves 90% with only a few hundred training epochs. We find that the combination of segmentation and classification networks in a single pipeline yields performance comparable to existing computational pipelines, while providing the additional benefit of an end-to-end, fully-automated analysis toolbox for ciliary motion.

Deep learning defines a new data-driven programming paradigm that constructs the internal system logic of a crafted neuron network through a set of training data. Deep learning (DL) has been widely adopted in many safety-critical scenarios. However, a plethora of studies have shown that the state-of-the-art DL systems suffer from various vulnerabilities which can lead to severe consequences when applied to real-world applications. Currently, the robustness of a DL system against adversarial attacks is usually measured by the accuracy of test data. Considering the limitation of accessible test data, good performance on test data can hardly guarantee the robustness and generality of DL systems. Different from traditional software systems which have clear and controllable logic and functionality, a DL system is trained with data and lacks thorough understanding. This makes it difficult for system analysis and defect detection, which could potentially hinder its real-world deployment without safety guarantees. In this paper, we propose DeepGauge, a comprehensive and multi-granularity testing criteria for DL systems, which renders a complete and multi-faceted portrayal of the testbed. The in-depth evaluation of our proposed testing criteria is demonstrated on two well-known datasets, five DL systems, with four state-of-the-art adversarial data generation techniques. The effectiveness of DeepGauge sheds light on the construction of robust DL systems.

Issues regarding explainable AI involve four components: users, laws & regulations, explanations and algorithms. Together these components provide a context in which explanation methods can be evaluated regarding their adequacy. The goal of this chapter is to bridge the gap between expert users and lay users. Different kinds of users are identified and their concerns revealed, relevant statements from the General Data Protection Regulation are analyzed in the context of Deep Neural Networks (DNNs), a taxonomy for the classification of existing explanation methods is introduced, and finally, the various classes of explanation methods are analyzed to verify if user concerns are justified. Overall, it is clear that (visual) explanations can be given about various aspects of the influence of the input on the output. However, it is noted that explanation methods or interfaces for lay users are missing and we speculate which criteria these methods / interfaces should satisfy. Finally it is noted that two important concerns are difficult to address with explanation methods: the concern about bias in datasets that leads to biased DNNs, as well as the suspicion about unfair outcomes.

MLtuner automatically tunes settings for training tunables (such as the learning rate, the momentum, the mini-batch size, and the data staleness bound) that have a significant impact on large-scale machine learning (ML) performance. Traditionally, these tunables are set manually, which is unsurprisingly error-prone and difficult to do without extensive domain knowledge. MLtuner uses efficient snapshotting, branching, and optimization-guided online trial-and-error to find good initial settings as well as to re-tune settings during execution. Experiments show that MLtuner can robustly find and re-tune tunable settings for a variety of ML applications, including image classification (for 3 models and 2 datasets), video classification, and matrix factorization. Compared to state-of-the-art ML auto-tuning approaches, MLtuner is more robust for large problems and over an order of magnitude faster.

The rapid development of deep learning methods has permitted the fast and accurate medical decision making from complex structured data, like CT images or MRI. However, some problems still exist in such applications that may lead to imperfect predictions. Previous observations have shown that, confounding factors, if handled inappropriately, will lead to biased prediction results towards some major properties of the data distribution. In other words, naïvely applying deep learning methods in these applications will lead to unfair prediction results for the minority group defined by the characteristics including age, gender, or even the hospital that collects the data, etc. In this paper, extending previous successes in correcting confounders, we propose a more stable method, namely Confounder Filtering, that can effectively reduce the influence of confounding factors, leading to better generalizability of trained discriminative deep neural networks, therefore, fairer prediction results. Our experimental results indicate that the Confounder Filtering method is able to improve the performance for different neural networks including CNN, LSTM, and other arbitrary architecture, different data types including CTscan, MRI, and EEG brain wave data, as well as different confounding factors including age, gender, and physical factors of medical devices etc.

Gaussian processes are a natural extension of the ubiquitous kernel regression to the Bayesian setting where the regression parameters are modelled as random variables with a Gaussian prior distribution [1]. Given the training observations, Gaussian processes generate posterior probabilities of the target or output for new inputs or observations, as a function of the training data and the input kernel function [2]. Gaussian process models and its variants have been applied in a number of diverse fields such as model predictive control and system analysis [3]-[7], latent variable models [8]-[11], multi-task learning [10], [12], [13], image analysis and synthesis [14]- [17], speech processing [18]-[20], and magnetic resonance imaging (MRI) [21], [22]. Gaussian processes have also been extended to a non-stationary regression setting [23]-[25] and for regression over complex-valued data [26]. Recently, Gaussian processes were shown to be useful in training and analysis of deep neural networks, and that a Gaussian process can be viewed as a neural network with a single infinite-dimensional layer of hidden units [27], [28].

In this paper, we propose variance networks, a new model that stores the learned information in the variances of the network weights. Surprisingly, no information gets stored in the expectations of the weights, therefore if we replace these weights with their expectations, we would obtain a random guess quality prediction. We provide a numerical criterion that uses the loss curvature to determine which random variables can be replaced with their expected values, and find that only a small fraction of weights is needed for ensembling. Variance networks represent a diverse ensemble that is more robust to adversarial attacks than conventional low-variance ensembles. The success of this model raises several counter-intuitive implications for the training and application of Deep Learning models.

Neural networks are commonly used as models for classification for a wide variety of tasks. Typically, a learned affine transformation is placed at the end of such models, yielding a per-class value used for classification. This classifier can have a vast number of parameters, which grows linearly with the number of possible classes, thus requiring increasingly more resources. In this work we argue that this classifier can be fixed, up to a global scale constant, with little or no loss of accuracy for most tasks, allowing memory and computational benefits. Moreover, we show that by initializing the classifier with a Hadamard matrix we can speed up inference as well. We discuss the implications for current understanding of neural network models.

Magnetic resonance imaging (MRI) has been proposed as a complimentary method to measure bone quality and assess fracture risk. However, manual segmentation of MR images of bone is time-consuming, limiting the use of MRI measurements in the clinical practice. The purpose of this paper is to present an automatic proximal femur segmentation method that is based on deep convolutional neural networks (CNNs). This study had institutional review board approval and written informed consent was obtained from all subjects. A dataset of volumetric structural MR images of the proximal femur from 86 subject were manually-segmented by an expert. We performed experiments by training two different CNN architectures with multiple number of initial feature maps and layers, and tested their segmentation performance against the gold standard of manual segmentations using fourfold cross-validation. Automatic segmentation of the proximal femur achieved a high dice similarity score of 0.94 0.05 with precision 0.95 0.02, and recall 0.94 0.08 using a CNN architecture based on 3D convolution exceeding the performance of 2D CNNs. The high segmentation accuracy provided by CNNs has the potential to help bring the use of structural MRI measurements of bone quality into clinical practice for management of osteoporosis.