Advances in deep neural networks (DNNs) and the availability of massive real-world data have enabled superhuman levels of accuracy on many AI tasks and ushered the explosive growth of AI workloads across the spectrum of computing devices. However, their superior accuracy comes at a high computational cost, which necessitates approaches beyond traditional computing paradigms to improve their operational efficiency. Leveraging the application-level insight of error resilience, we demonstrate how approximate computing (AxC) can significantly boost the efficiency of AI platforms and play a pivotal role in the broader adoption of AI-based applications and services. To this end, we present RaPiD, a multi-tera operations per second (TOPS) AI hardware accelerator core (fabricated at 14-nm technology) that we built from the ground-up using AxC techniques across the stack including algorithms, architecture, programmability, and hardware. We highlight the workload-guided systematic explorations of AxC techniques for AI, including custom number representations, quantization/pruning methodologies, mixed-precision architecture design, instruction sets, and compiler technologies with quality programmability, employed in the RaPiD accelerator.

With the passage of recent federal legislation, many medical institutions are now responsible for reaching target hospital readmission rates. Chronic diseases account for many hospital readmissions and chronic obstructive pulmonary disease has been recently added to the list of diseases for which the United States government penalizes hospitals incurring excessive readmissions. Though there have been efforts to statistically predict those most in danger of readmission, a few have focused primarily on unstructured clinical notes. We have proposed a framework, which uses natural language processing to analyze clinical notes and predict readmission. Many algorithms within the field of data mining and machine learning exist, so a framework for component selection is created to select the best components. Naïve Bayes using Chi-Squared feature selection offers an AUC of 0.690 while maintaining fast computational times.

Care coordination (CC) is a critical function in the healthcare system and is currently a national priority in the US. Improved CC help reduce avoidable hospital admissions and readmissions. A CC system is a common platform where in all the care providers including patients interact with each other thereby sharing a common set of information and knowledge about a patient's condition that will result in improved quality of care. This study reviews various frameworks and strategies in CC, and finds that most CC systems do not address the social and community service coordination dimensions of CC. This is especially important as there is an increasing move towards “aging at home” in the country. So this study also presents the design and implementation of a community-based CC system for elder-care. Specifically, the approach adopted in this study was to develop a Learning Community for service providers, so better communication and learning among providers would result in increased care coordination across healthcare, social, and community eco-systems in the area of elder-care.

Unplanned hospital readmission is a costly problem in the United States. Patients treated and readmitted within 30 days cost tax payers up to $26 billion annually. In 2013 the U.S. federal government began to reduce payments to hospitals with excessive patient readmissions. Predictive modeling using machine learning can be a useful tool to help identify patients most likely to need readmission. However, current systems have several shortcomings. When creating predictive models for hospital readmission, existing methods either build models using data from a single hospital or naively combining data from multiple hospitals. Because hospitals often have different data distributions, models created from a single hospital's data are often biased. Additionally, models created from combined data overlook local data distributions. In this paper, we propose, LTEL, which uses an ensemble of topic specific models to leverage data from multiple hospitals. LTEL creates models based on latent topics derived from different hospitals. Models are built and evaluated incorporating federal financial penalties. The dataset contains data collected from 16 regional hospitals. Compared to baseline methods, LTEL significantly outperforms the best performing baseline method for cost reduction.

Natural language processing is a field of computer science, which focuses on interactions between computers and human (natural) languages. The human languages are ambiguous unlike Computer languages, which make its analysis and processing difficult. Most of the data present these days is in unstructured form (such as: Accident reports, Patient discharge summary, Criminal records etc), which makes it hard for computers to understand for further use and analysis. This unstructured text needs to be converted into structured form by clearly defining the sentence boundaries, word boundaries and context dependent character boundaries for further analysis. This paper proposes a component-based domain-independent text analysis system for processing of the natural language known as Domain-independent Natural Language Processing System (DINLP). Further the paper discusses the system capability and its application in the area of bioinformatics through the case study.

There is a significant increase in attention being paid to personal wellness as a preventative strategy in healthcare. At the same time, chronic diseases are the major cause of mortality, accounting for 7 out of 10 deaths in the United States. Healthcare costs involved in managing chronic diseases are also very high. So there is a need to help better maintain individual wellness, as well as better manage chronic conditions. Predictive analytics based clinical decision support systems need to be developed to help individuals and healthcare providers to better manage wellness or chronic conditions. In this paper, we investigate two different classifiers to predict the wellness outcome and the occurrence of a chronic condition (diabetes). The models were evaluated on the basis of overall accuracy, root mean squared error and Area under ROC. National CDC-NHANES data that is based on the health and nutritional status of individuals in the United States is used to develop the models.

The prevalence of end-stage renal disease in the U.S. Has grown significantly, and continues to do so. Organ transplantation generally has better overall patient outcomes than dialysis. But there is a significant shortage of kidneys. This shortage is exacerbated by the need for kidneys for patients with dual organ transplantations. So the kidney allocation problem is a significant challenge. Predictive analytics based clinical decision support systems need to be developed to help physicians make difficult organ allocation decisions. In this paper, we investigate two different classifiers to predict the outcomes of kidney-liver dual transplant patients. The models were evaluated on the basis of overall accuracy, root mean squared error and Area under ROC. UNOS data was used to develop the models.

We propose an algorithm that shows improvement in computation timeof existing image encryption-decryption methods. The algorithm also introduces additional one-level of encryption key into the existing methods. Image compression properties of the DWT (Discrete Wavelet Transform) are utilized to bring the improvement. The new algorithms retain MSE between input and decrypted images and the robustness of existing algorithms.

Stereo matching algorithms conventionally match over a range of disparities sufficient to encompass all visible 3D scene points. Human vision, however, works over a narrow band of disparities-Panum's fusional band-whose typical range may be as little as 1/20 of the full range of disparities for visible points. Only points inside the band are fused visually; the remainder of points are seen diplopically. A probabilistic approach is presented for dense stereo matching under the Panum band restriction. It is shown that existing dense stereo algorithms are inadequate in this problem setting and the main problem is segmentation, marking the image into the areas that fall inside the band. An approximation is derived that makes up for missing out-of-band information with a "proxy" based on image autocorrelation. It is shown that the Panum Proxy algorithm achieves accuracy close to what can be obtained when the full disparity band is available, and with gains of between one and two orders of magnitude in computation time. There are also substantial gains in computation space. Panum band processing is also demonstrated in an active stereopsis framework.

A new method for localising and recognising hand poses and objects in real-time is presented. This problem is important in vision-driven applications where it is natural for a user to combine hand gestures and real objects when interacting with a machine. Examples include using a real eraser to remove words from a document displayed on an electronic surface. In this paper the task of simultaneously recognising object classes, hand gestures and detecting touch events is cast as a single classification problem. A random forest algorithm is employed which adaptively selects and combines a minimal set of appearance, shape and stereo features to achieve maximum class discrimination for a given image. This minimal set leads to both efficiency at run time and good generalisation. Unlike previous stereo works which explicitly construct disparity maps, here the stereo matching costs are used directly as visual cue and only computed on-demand, i.e. only for pixels where they are necessary for recognition. This leads to improved efficiency. The proposed method is assessed on a database of a variety of objects and hand poses selected for interacting on a flat surface in an office environment.

Stereo matching algorithms conventionally match over a range of disparities sufficient to encompass all visible 3D scene points. Human vision however does not do this. It works over a narrow band of disparities - Panum’s fusional band - whose typical range may be as little as 1/20 of the full range of disparities for visible points. Points inside the band are fused visually and the remainder of points are seen as "diplopic" - that is with double vision. The Panum band restriction is important also in machine vision, both with active (pan/tilt) cameras, and with high resolution cameras and digital pan/tilt.

We describe a learning-based method for recovering 3D human body pose from single images and monocular image sequences. Our approach requires neither an explicit body model nor prior labeling of body parts in the image. Instead, it recovers pose by direct nonlinear regression against shape descriptor vectors extracted automatically from image silhouettes. For robustness against local silhouette segmentation errors, silhouette shape is encoded by histogram-of-shape-contexts descriptors. We evaluate several different regression methods: ridge regression, Relevance Vector Machine (RVM) regression, and Support Vector Machine (SVM) regression over both linear and kernel bases. The RVMs provide much sparser regressors without compromising performance, and kernel bases give a small but worthwhile improvement in performance. The loss of depth and limb labeling information often makes the recovery of 3D pose from single silhouettes ambiguous. To handle this, the method is embedded in a novel regressive tracking framework, using dynamics from the previous state estimate together with a learned regression value to disambiguate the pose. We show that the resulting system tracks long sequences stably. For realism and good generalization over a wide range of viewpoints, we train the regressors on images resynthesized from real human motion capture data. The method is demonstrated for several representations of full body pose, both quantitatively on independent but similar test data and qualitatively on real image sequences. Mean angular errors of 4-6 degrees are obtained for a variety of walking motions.

We describe a sparse Bayesian regression method for recovering 3D human body motion directly from silhouettes extracted from monocular video sequences. No detailed body shape model is needed, and realism is ensured by training on real human motion capture data. The tracker estimates 3D body pose by using Relevance Vector Machine regression to combine a learned autoregressive dynamical model with robust shape descriptors extracted automatically from image silhouettes. We studied several different combination methods, the most effective being to learn a nonlinear observation-update correction based on joint regression with respect to the predicted state and the observations. We demonstrate the method on a 54-parameter full body pose model, both quantitatively using motion capture based test sequences, and qualitatively on a test video sequence.