Frequently implicated in psychotic spectrum disorders, the amygdala serves as an important hub for elucidating the convergent and divergent neural substrates in schizophrenia and bipolar disorder, the two most studied groups of psychotic spectrum conditions. A systematic search of electronic databases through December 2017 was conducted to identify neuroimaging studies of the amygdala in schizophrenia and bipolar disorder, focusing on structural MRI, diffusion tensor imaging (DTI), and resting-state functional connectivity studies, with an emphasis on cross-diagnostic studies. Ninety-four independent studies were selected for the present review (49 structural MRI, 27 DTI, and 18 resting-state functional MRI studies). Also selected, and analyzed in a separate meta-analysis, were 33 volumetric studies with the amygdala as the region-of-interest. Reduced left, right, and total amygdala volumes were found in schizophrenia, relative to both healthy controls and bipolar subjects, even when restricted to cohorts in the early stages of illness. No volume abnormalities were observed in bipolar subjects relative to healthy controls. Shape morphometry studies showed either amygdala deformity or no differences in schizophrenia, and no abnormalities in bipolar disorder. In contrast to the volumetric findings, DTI studies of the uncinate fasciculus tract (connecting the amygdala with the medial- and orbitofrontal cortices) largely showed reduced fractional anisotropy (a marker of white matter microstructure abnormality) in both schizophrenia and bipolar patients, with no cross-diagnostic differences. While decreased amygdalar-orbitofrontal functional connectivity was generally observed in schizophrenia, varying patterns of amygdalar-orbitofrontal connectivity in bipolar disorder were found. Future studies can consider adopting longitudinal approaches with multimodal imaging and more extensive clinical subtyping to probe amygdalar subregional changes and their relationship to the sequelae of psychotic disorders.

No abstract

No abstract

A GPU's computing power lies in its abundant memory bandwidth and massive parallelism. However, its hardware thread schedulers, despite being able to quickly distribute computation to processors, often fail to capitalize on program characteristics effectively, achieving only a fraction of the GPU's full potential. Moreover, current GPUs do not allow programmers or compilers to control this thread scheduling, forfeiting important optimization opportunities at the program level. This paper presents a transformation centered on Streaming Multiprocessors (SM); this software approach to circumventing the limitations of the hardware scheduler allows flexible program-level control of scheduling. By permitting precise control of job locality on SMs, the transformation overcomes inherent limitations in prior methods.With this technique, flexible control of GPU scheduling at the program level becomes feasible, which opens up new opportunities for GPU program optimizations. The second part of the paper explores how the new opportunities could be leveraged for GPU performance enhancement, what complexities there are, and how to address them. We show that some simple optimization techniques can enhance co-runs of multiple kernels and improve data locality of irregular applications, producing 20-33% average increase in performance, system throughput, and average turnaround time.

No abstract

This paper explores solutions for enabling efficient supports of position independence of pointer-based data structures on byteaddressable None-Volatile Memory (NVM). When a dynamic data structure (e.g., a linked list) gets loaded from persistent storage into main memory in different executions, the locations of the elements contained in the data structure could differ in the address spaces from one run to another. As a result, some special support must be provided to ensure that the pointers contained in the data structures always point to the correct locations, which is called position independence. This paper shows the insufficiency of traditional methods in supporting position independence on NVM. It proposes a concept called implicit self-contained representations of pointers, and develops two such representations named off-holder and Region ID in Value (RIV) to materialize the concept. Experiments show that the enabled representations provide much more efficient and flexible support of position independence for dynamic data structures, alleviating a major issue for effective data reuses on NVM. CCS CONCEPTS • Hardware ! Memory and dense storage; • Computer systems organization ! Architectures; • Software and its engineering ! Compilers; General programming languages;

No abstract

Recently smoothing deep neural network based classifiers via isotropic Gaussian perturbation is shown to be an effective and scalable way to provide state-of-the-art probabilistic robustness guarantee against ℓ2 norm bounded adversarial perturbations. However, how to train a good base classifier that is accurate and robust when smoothed has not been fully investigated. In this work, we derive a new regularized risk, in which the regularizer can adaptively encourage the accuracy and robustness of the smoothed counterpart when training the base classifier. It is computationally efficient and can be implemented in parallel with other empirical defense methods. We discuss how to implement it under both standard (non-adversarial) and adversarial training scheme. At the same time, we also design a new certification algorithm, which can leverage the regularization effect to provide tighter robustness lower bound that holds with high probability. Our extensive experimentation demonstrates the effectiveness of the proposed training and certification approaches on CIFAR-10 and ImageNet datasets.