The research and development of neural networks is flourishing thanks to recent advancements in computational power, the discovery of new algorithms, and an increase in labelled data. Before the current explosion of activity in the space, the practical applications of neural networks were limited. Much of the recent research has allowed for broad application, the heavy computational requirements for machine learning models still restrain it from truly entering the mainstream. Now, emerging algorithms are on the cusp of pushing neural networks into more conventional applications through exponentially increased efficiency. Neural networks are a prominent focal point in the current state of computer science research. They are inspired by complex human biology, which, for all but the most niche use cases, still outperforms computers on most conceivable scales.

Despite the plethora of marketing hype touting "universal" deep learning models that can be magically applied to any domain and any question without requiring additional training, the reality is that today's deep learning systems are largely bespoke creations, custom built for each individual application. Transfer learning, automated neural architecture selection and automatic tuning can reduce much of the manual labor required in constructing and tuning the algorithm itself and minimize the amount of novel training data required, but at the end of the day, every new deep learning application requires the creation of a new custom-built model. It is one of the great ironies of the deep learning revolution that despite the biggest advances today increasingly coming from machine-assisted development, humans are still needed to create and curate the vast training datasets required and to oversee the development process. While there are an increasing number of point-and-click model generators, they still require labeled training data and still require that a new model be created for each application. Moreover, transfer learning is still extremely limited, encoding only low-level pattern relationship knowledge rather than the high-level abstract knowledge representation and transfer that humans make use of when learning new tasks.

Autonomous driving, healthcare or retail are just some of the areas where Computer Vision has allowed us to achieve things that, until recently, were considered impossible. Today the dream of a self driving car or automated grocery store does not sound so futuristic anymore. In fact, we are using Computer Vision every day -- when we unlock the phone with our face or automatically retouch photos before posting them on social media. Convolutional Neural Networks are possibly the most crucial building blocks behind this huge successes. This time we are going to broaden our understanding of how neural networks work with ideas specific to CNNs.

First, we import the necessary modules. We have 12 different classes with 10,234 training images and 2,558 validation images. For those with NVIDIA RTX graphic cards, mixed precision greatly speeds up training and halves the memory requirement. The results are already looking pretty good even without training the top layers. We will now unfreeze the top layers and find a suitable learning rate.

In [1, 2], we have explored the theoretical aspects of feature extraction optimization processes for solving largescale problems and overcoming machine learning limitations. Majority of optimization algorithms that have been introduced in [1, 2] guarantee the optimal performance of supervised learning, given offline and discrete data, to deal with curse of dimensionality (CoD) problem. These algorithms, however, are not tailored for solving emerging learning problems. One of the important issues caused by online data is lack of sufficient samples per class. Further, traditional machine learning algorithms cannot achieve accurate training based on limited distributed data, as data has proliferated and dispersed significantly. Machine learning employs a strict model or embedded engine to train and predict which still fails to learn unseen classes and sufficiently use online data. In this chapter, we introduce these challenges elaborately. We further investigate Meta-Learning (MTL) algorithm, and their application and promises to solve the emerging problems by answering how autonomous agents can learn to learn?.

Breast cancer is the most common cancer and is the leading cause of cancer death among women worldwide. Detection of breast cancer, while it is still small and confined to the breast, provides the best chance of effective treatment. Computer Aided Detection (CAD) systems that detect cancer from mammograms will help in reducing the human errors that lead to missing breast carcinoma. Literature is rich of scientific papers for methods of CAD design, yet with no complete system architecture to deploy those methods. On the other hand, commercial CADs are developed and deployed only to vendors' mammography machines with no availability to public access. This paper presents a complete CAD; it is complete since it combines, on a hand, the rigor of algorithm design and assessment (method), and, on the other hand, the implementation and deployment of a system architecture for public accessibility (system). (1) We develop a novel algorithm for image enhancement so that mammograms acquired from any digital mammography machine look qualitatively of the same clarity to radiologists' inspection; and is quantitatively standardized for the detection algorithms. (2) We develop novel algorithms for masses and microcalcifications detection with accuracy superior to both literature results and the majority of approved commercial systems. (3) We design, implement, and deploy a system architecture that is computationally effective to allow for deploying these algorithms to cloud for public access.

Non-exhaustive learning (NEL) is an emerging machine-learning paradigm designed to confront the challenge of non-stationary environments characterized by anon-exhaustive training sets lacking full information about the available classes.Unlike traditional supervised learning that relies on fixed models, NEL utilizes self-adjusting machine learning to better accommodate the non-stationary nature of the real-world problem, which is at the root of many recently discovered limitations of deep learning. Some of these hurdles led to a surge of interest in several research areas relevant to NEL such as open set classification or zero-shot learning. The presented study which has been motivated by two important applications proposes a NEL algorithm built on a highly flexible, doubly non-parametric Bayesian Gaussian mixture model that can grow arbitrarily large in terms of the number of classes and their components. We report several experiments that demonstrate the promising performance of the introduced model for NEL.

Deep neural network (DNN) regression models are widely used in applications requiring state-of-the-art predictive accuracy. However, until recently there has been little work on accurate uncertainty quantification for predictions from such models. We add to this literature by outlining an approach to constructing predictive distributions that are `marginally calibrated'. This is where the long run average of the predictive distributions of the response variable matches the observed empirical margin. Our approach considers a DNN regression with a conditionally Gaussian prior for the final layer weights, from which an implicit copula process on the feature space is extracted. This copula process is combined with a non-parametrically estimated marginal distribution for the response. The end result is a scalable distributional DNN regression method with marginally calibrated predictions, and our work complements existing methods for probability calibration. The approach is first illustrated using two applications of dense layer feed-forward neural networks. However, our main motivating applications are in likelihood-free inference, where distributional deep regression is used to estimate marginal posterior distributions. In two complex ecological time series examples we employ the implicit copulas of convolutional networks, and show that marginal calibration results in improved uncertainty quantification. Our approach also avoids the need for manual specification of summary statistics, a requirement that is burdensome for users and typical of competing likelihood-free inference methods.

To accurately analyze changes of anatomical structures in longitudinal imaging studies, consistent segmentation across multiple time-points is required. Existing solutions often involve independent registration and segmentation components. Registration between time-points is used either as a prior for segmentation in a subsequent time point or to perform segmentation in a common space. In this work, we propose a novel hybrid convolutional neural network (CNN) that integrates segmentation and registration into a single procedure. We hypothesize that the joint optimization leads to increased performance on both tasks. The hybrid CNN is trained by minimizing an integrated loss function composed of four different terms, measuring segmentation accuracy, similarity between registered images, deformation field smoothness, and segmentation consistency. We applied this method to the segmentation of white matter tracts, describing functionally grouped axonal fibers, using N=8045 longitudinal brain MRI data of 3249 individuals. The proposed method was compared with two multistage pipelines using two existing segmentation methods combined with a conventional deformable registration algorithm. In addition, we assessed the added value of the joint optimization for segmentation and registration separately. The hybrid CNN yielded significantly higher accuracy, consistency and reproducibility of segmentation than the multistage pipelines, and was orders of magnitude faster. Therefore, we expect it can serve as a novel tool to support clinical and epidemiological analyses on understanding microstructural brain changes over time.

Tract-specific diffusion measures, as derived from brain diffusion MRI, have been linked to white matter tract structural integrity and neurodegeneration. As a consequence, there is a large interest in the automatic segmentation of white matter tract in diffusion tensor MRI data. Methods based on the tractography are popular for white matter tract segmentation. However, because of the limited consistency and long processing time, such methods may not be suitable for clinical practice. We therefore developed a novel convolutional neural network based method to directly segment white matter tract trained on a low-resolution dataset of 9149 DTI images. The method is optimized on input, loss function and network architecture selections. We evaluated both segmentation accuracy and reproducibility, and reproducibility of determining tract-specific diffusion measures. The reproducibility of the method is higher than that of the reference standard and the determined diffusion measures are consistent. Therefore, we expect our method to be applicable in clinical practice and in longitudinal analysis of white matter microstructure.