Parametric generative deep models are state-of-the-art for photo and non-photo realistic image stylization. However, learning complicated image representations requires computeintense modelsparametrized by a huge number of weights, which in turn requires large datasets to make learning successful. Nonparametric exemplar-based generation is a technique that works well to reproduce style from small datasets, but is also computeintensive. Theseaspects are a drawback for the practice of digital AI artists: typically one wants to use a small set of stylization images, and needs a fast flexible model in order to experiment with it. With this motivation, our work has these contributions: (i) a novel stylization method called Fully Adversarial Mosaics (FAMOS) that combines the strengths of both parametric and nonparametric approaches; (ii) multiple ablations and image examples that analyze the method and show its capabilities; (iii) source code that will empower artists and machine learning researchers to use and modify FAMOS. Tiling of small stones was a classical ancient art form, and in modern times there are efficient algorithms to produce such mosaics (with non-overlapping tiles) digitally [10]. Seamless mosaics in the style of the Renaissance painter Archimboldo are more challenging, but modern deep learning methods allow efficient seamless image stylization. Neural style transfer [4] uses filter statistics (pretrained on a huge dataset) of a style image to optimize an output image.

Learning the activities of animals is important for the purpose of monitoring their welfare vis a vis their behaviour with respect to their environment and conspecifics. While previous works have largely focused on activity recognition in a single animal, little or no work has been done in learning the collective behaviour of animals. In this work, we address the problem of recognising the collective movement activities of a group of sheep in a flock. We present a discriminative framework that learns to track the positions and velocities of all the animals in the flock in an online manner whilst estimating their collective activity. We investigate the performance of two simple deep network architectures and show that we can learn the collective activities with good accuracy even when the distribution of the activities is skewed.

Random projections are able to perform dimension reduction efficiently for datasets with nonlinear low-dimensional structures. One well-known example is that random matrices embed sparse vectors into a low-dimensional subspace nearly isometrically, known as the restricted isometric property in compressed sensing. In this paper, we explore some applications of random projections in deep neural networks. We provide the expressive power of fully connected neural networks when the input data are sparse vectors or form a low-dimensional smooth manifold. We prove that the number of neurons required for approximating a Lipschitz function with a prescribed precision depends on the sparsity or the dimension of the manifold and weakly on the dimension of the input vector. The key in our proof is that random projections embed stably the set of sparse vectors or a low-dimensional smooth manifold into a low-dimensional subspace. Based on this fact, we also propose some new neural network models, where at each layer the input is first projected onto a low-dimensional subspace by a random projection and then the standard linear connection and non-linear activation are applied. In this way, the number of parameters in neural networks is significantly reduced, and therefore the training of neural networks can be accelerated without too much performance loss.

We propose a new architecture and training methodology for generative adversarial networks. Current approaches attempt to learn the transformation from a noise sample to a generated data sample in one shot. Our proposed generator architecture, called $\textit{ChainGAN}$, uses a two-step process. It first attempts to transform a noise vector into a crude sample, similar to a traditional generator. Next, a chain of networks, called $\textit{editors}$, attempt to sequentially enhance this sample. We train each of these units independently, instead of with end-to-end backpropagation on the entire chain. Our model is robust, efficient, and flexible as we can apply it to various network architectures. We provide rationale for our choices and experimentally evaluate our model, achieving competitive results on several datasets.

The training method of repetitively feeding all samples into a pre-defined network for image classification has been widely adopted by current state-of-the-art. In this work, we provide a new method, which can be leveraged to train classification networks in a more efficient way. Starting with a warm-up step, we propose to continually repeat a Drop-and-Pick (DaP) learning strategy. In particular, we drop those easy samples to encourage the network to focus on studying hard ones. Meanwhile, by picking up all samples periodically during training, we aim to recall the memory of the networks to prevent catastrophic forgetting of previously learned knowledge. Our DaP learning method can recover 99.88%, 99.60%, 99.83% top-1 accuracy on ImageNet for ResNet-50, DenseNet-121, and MobileNet-V1 but only requires 75% computation in training compared to those using the classic training schedule. Furthermore, our pre-trained models are equipped with strong knowledge transferability when used for downstream tasks, especially for hard cases. Extensive experiments on object detection, instance segmentation and pose estimation can well demonstrate the effectiveness of our DaP training method.

The analysis of single CT slices extracted at the third lumbar vertebra (L3) has garnered significant clinical interest in the past few years, in particular in regards to quantifying sarcopenia (muscle loss). In this paper, we propose an efficient method to automatically detect the L3 slice in 3D CT images. Our method works with images with a variety of fields of view, occlusions, and slice thicknesses. 3D CT images are first converted into 2D via Maximal Intensity Projection (MIP), reducing the dimensionality of the problem. The MIP images are then used as input to a 2D fully-convolutional network to predict the L3 slice locations in the form of 2D confidence maps. In addition we propose a variant architecture with less parameters allowing 1D confidence map prediction and slightly faster prediction time without loss of accuracy. Quantitative evaluation of our method on a dataset of 1006 3D CT images yields a median error of 1mm, similar to the inter-rater median error of 1mm obtained from two annotators, demonstrating the effectiveness of our method in efficiently and accurately detecting the L3 slice.

Several fundamental changes in technology indicate domain-specific hardware and software co-design is the only path left. In this context, architecture, system, data management, and machine learning communities pay greater attention to innovative big data and AI algorithms, architecture, and systems. Unfortunately, complexity, diversity, frequently-changed workloads, and rapid evolution of big data and AI systems raise great challenges. First, the traditional benchmarking methodology that creates a new benchmark or proxy for every possible workload is not scalable, or even impossible for Big Data and AI benchmarking. Second, it is prohibitively expensive to tailor the architecture to characteristics of one or more application or even a domain of applications. We consider each big data and AI workload as a pipeline of one or more classes of units of computation performed on different initial or intermediate data inputs, each class of which we call a data motif. On the basis of our previous work that identifies eight data motifs taking up most of the run time of a wide variety of big data and AI workloads, we propose a scalable benchmarking methodology that uses the combination of one or more data motifs---to represent diversity of big data and AI workloads. Following this methodology, we present a unified big data and AI benchmark suite---BigDataBench 4.0, publicly available from~\url{http://prof.ict.ac.cn/BigDataBench}. This unified benchmark suite sheds new light on domain-specific hardware and software co-design: tailoring the system and architecture to characteristics of the unified eight data motifs other than one or more application case by case. Also, for the first time, we comprehensively characterize the CPU pipeline efficiency using the benchmarks of seven workload types in BigDataBench 4.0.

Most healthcare systems and insurance providers know who the high-risk patients are: Usually people living with chronic conditions--diabetes, asthma, heart disease--who aren't managing their health well enough. These individuals are at a high risk of ending up in the ER. There are several problems with this high-risk list approach to caring for patients. First, these risk scores are often compiled using insurance claims. This is a limited data source, and it is all about the past, not the present.

Health Care systems around the world are facing a fundamental financial crisis. Moreover, there is the issue of misdiagnosis or time taken to correctly diagnose due to information overload across the system. In part these failures are due to over worked medical staff and in part due to handling Big Data with inadequate information management systems. There is a real cost to both misdiagnosis and delays for correct diagnosis. These costs are both economic and also personal as people have to deal with the consequences of shattered lives that may result.