• ### Generalized Convolutional Neural Networks for Point Cloud Data(1707.06719)

Oct. 19, 2018 cs.CV, stat.ML
The introduction of cheap RGB-D cameras, stereo cameras, and LIDAR devices has given the computer vision community 3D information that conventional RGB cameras cannot provide. This data is often stored as a point cloud. In this paper, we present a novel method to apply the concept of convolutional neural networks to this type of data. By creating a mapping of nearest neighbors in a dataset, and individually applying weights to spatial relationships between points, we achieve an architecture that works directly with point clouds, but closely resembles a convolutional neural net in both design and behavior. Such a method bypasses the need for extensive feature engineering, while proving to be computationally efficient and requiring few parameters.
• ### Parallel local approximation MCMC for expensive models(1607.02788)

Dec. 25, 2017 stat.CO, stat.ME, stat.AP
Performing Bayesian inference via Markov chain Monte Carlo (MCMC) can be exceedingly expensive when posterior evaluations invoke the evaluation of a computationally expensive model, such as a system of partial differential equations. In recent work [Conrad et al. JASA 2016, arXiv:1402.1694], we described a framework for constructing and refining local approximations of such models during an MCMC simulation. These posterior--adapted approximations harness regularity of the model to reduce the computational cost of inference while preserving asymptotic exactness of the Markov chain. Here we describe two extensions of that work. First, we prove that samplers running in parallel can collaboratively construct a shared posterior approximation while ensuring ergodicity of each associated chain, providing a novel opportunity for exploiting parallel computation in MCMC. Second, focusing on the Metropolis--adjusted Langevin algorithm, we describe how a proposal distribution can successfully employ gradients and other relevant information extracted from the approximation. We investigate the practical performance of our strategies using two challenging inference problems, the first in subsurface hydrology and the second in glaciology. Using local approximations constructed via parallel chains, we successfully reduce the run time needed to characterize the posterior distributions in these problems from days to hours and from months to days, respectively, dramatically improving the tractability of Bayesian inference.
• ### Neutron shielding and activation of the MASTU device and surrounds(1412.0860)

Dec. 2, 2014 physics.ins-det
A significant functional upgrade is planned for the Mega Ampere Spherical Tokamak (MAST) device, located at Culham in the UK, including the implementation of a notably greater neutral beam injection power. This upgrade will cause the emission of a substantially increased intensity of neutron radiation for a substantially increased amount of time upon operation of the device. Existing shielding and activation precautions are shown to prove insufficient in some regards, and recommendations for improvements are made, including the following areas: shielding doors to MAST shielded facility enclosure (known as "the blockhouse"); north access tunnel; blockhouse roof; west cabling duct. In addition, some specific neutronic dose rate questions are addressed and answered; those discussed here relate to shielding penetrations and dose rate reflected from the air above the device ("skyshine").
• ### Low-Rank Approximations for Conditional Feedforward Computation in Deep Neural Networks(1312.4461)

Jan. 28, 2014 cs.LG
Scalability properties of deep neural networks raise key research questions, particularly as the problems considered become larger and more challenging. This paper expands on the idea of conditional computation introduced by Bengio, et. al., where the nodes of a deep network are augmented by a set of gating units that determine when a node should be calculated. By factorizing the weight matrix into a low-rank approximation, an estimation of the sign of the pre-nonlinearity activation can be efficiently obtained. For networks using rectified-linear hidden units, this implies that the computation of a hidden unit with an estimated negative pre-nonlinearity can be ommitted altogether, as its value will become zero when nonlinearity is applied. For sparse neural networks, this can result in considerable speed gains. Experimental results using the MNIST and SVHN data sets with a fully-connected deep neural network demonstrate the performance robustness of the proposed scheme with respect to the error introduced by the conditional computation process.
• ### Virialization of high redshift dark matter haloes(1007.4196)

June 3, 2011 astro-ph.CO
We present results of a study of the virial state of high redshift dark matter haloes in an N-body simulation. We find that the majority of collapsed, bound haloes are not virialized at any redshift slice in our study ($z=15-6$) and have excess kinetic energy. At these redshifts, merging is still rampant and the haloes cannot strictly be treated as isolated systems. To assess if this excess kinetic energy arises from the environment, we include the surface pressure term in the virial equation explicitly and relax the assumption that the density at the halo boundary is zero. Upon inclusion of the surface term, we find that the haloes are much closer to virialization, however, they still have some excess kinetic energy. We report trends of the virial ratio including the extra surface term with three key halo properties: spin, environment, and concentration. We find that haloes with closer neighbors depart more from virialization, and that haloes with larger spin parameters do as well. We conclude that except at the lowest masses ($M < 10^6 \Msun$), dark matter haloes at high redshift are not fully virialized. This finding has interesting implications for galaxy formation at these high redshifts, as the excess kinetic energy will impact the subsequent collapse of baryons and the formation of the first disks and/or baryonic structures.