By means of B-calculus [V. Garcia-Morales, Phys. Lett. A 376 (2012) 26451 a universal map for deterministic cellular automata (CAs) has been derived. The latter is shown here to be invariant upon certain transformatio...
详细信息
By means of B-calculus [V. Garcia-Morales, Phys. Lett. A 376 (2012) 26451 a universal map for deterministic cellular automata (CAs) has been derived. The latter is shown here to be invariant upon certain transformations (global complementation, reflection and shift). When constructing CA rules in terms of rules of lower range a new symmetry, "invariance under construction" is uncovered. Modular arithmetic is also reformulated within B-calculus and a new symmetry of certain totalistic CA rules, which calculate the Pascal simplices modulo an integer number p, is then also uncovered. (C) 2012 Elsevier B.V. All rights reserved.
This paper deals with the scattered deletion of a language by another language, i.e. with scattered residuals. We introduce the notion of the scattered syntactical monoid. The main result is a Myhill-Nerode-like theor...
详细信息
This paper deals with the scattered deletion of a language by another language, i.e. with scattered residuals. We introduce the notion of the scattered syntactical monoid. The main result is a Myhill-Nerode-like theorem for languages L with the property that the commutative closure of L, com(L), is a regular language. We investigate some properties of these families of languages related to the cardinality of the scattered syntactical monoid.
The spread of deep learning on embedded devices has prompted the development of numerous methods to optimize the deployment of deep neural networks (DNNs). Works have mainly focused on: 1) efficient DNN architectures;...
详细信息
The spread of deep learning on embedded devices has prompted the development of numerous methods to optimize the deployment of deep neural networks (DNNs). Works have mainly focused on: 1) efficient DNN architectures;2) network optimization techniques, such as pruning and quantization;3) optimized algorithms to speed up the execution of the most computational intensive layers;and 4) dedicated hardware to accelerate the data flow and computation. However, there is a lack of research on cross-level optimization as the space of approaches becomes too large to test and obtain a globally optimized solution. Thus, leading to suboptimal deployment in terms of latency, accuracy, and memory. In this work, we first detail and analyze the methods to improve the deployment of DNNs across the different levels of software optimization. Building on this knowledge, we present an automated exploration framework to ease the deployment of DNNs. The framework relies on a reinforcement learning search that, combined with a deep learning inference framework, automatically explores the design space and learns an optimized solution that speeds up the performance and reduces the memory on embedded CPU platforms. Thus, we present a set of results for state-of-the-art DNNs on a range of Arm Cortex-A CPU platforms achieving up to 4x improvement in performance and over 2x reduction in memory with negligible loss in accuracy with respect to the BLAS floating-point implementation.
In the situation calculus, it is sometimes necessary to prove that certain properties are true in all world states accessible from the initial state. This is the case for some forms of reasoning about the physical wor...
详细信息
In the situation calculus, it is sometimes necessary to prove that certain properties are true in all world states accessible from the initial state. This is the case for some forms of reasoning about the physical world, for certain planning applications, and for verifying integrity constraints in databases. Not surprisingly, this requires a suitable form of mathematical induction. This paper motivates the need for proving properties of states in the situation calculus, proposes appropriate induction principles for this task, and gives examples of their use in databases and for reasoning about the physical world.
SIGNIFICANCE Binocular summation is a sensitive metric of binocular integration. As such, characterization of the mechanisms underlying binocular summation is a key step in translating and applying this knowledge to a...
详细信息
SIGNIFICANCE Binocular summation is a sensitive metric of binocular integration. As such, characterization of the mechanisms underlying binocular summation is a key step in translating and applying this knowledge to abnormal binocular systems afflicted with strabismus and amblyopia. PURPOSE computational models of binocular summation have advocated the operation of mechanisms sensitive to the interocular phase disparity of first-order carrier gratings. This study investigated if such generalization depended on carrier spatial frequency and orientation. METHODS Monocular and binocular contrast detection thresholds were measured in nine observers with normal binocular vision. Stimuli comprised Gabor targets presented with one of three spatial frequencies (1, 3, and 9 cycles per degree [cpd]), two orientations (horizontal and vertical), and five interocular phase disparities (0, +/-pi/2, +/-pi radians). Horizontal and vertical fixation disparities were measured for each binocular threshold condition. Binocular summation ratios were computed by dividing the mean monocular detection threshold by the respective binocular detection threshold. RESULTS Binocular summation ratio varied significantly with interocular phase disparity for the 1- and 3-cpd horizontal and vertical gratings. Phase dependency was reduced with the 9-cpd horizontal grating and absent for the 9-cpd vertical grating, even though binocular summation ratio exceeded predictions of probability summation. computational modeling that incorporated the variability of fixation disparity into a vector summation model predicted a reduction in peak binocular summation ratio with increasing carrier spatial frequency but did not account for the reduction of phase sensitivity noted with the 9-cpd stimulus. CONCLUSIONS Binocular summation magnitude is less dependent on interocular phase disparities for carrier spatial frequencies that exceed 3 cpd, especially with vertical gratings. Although vergence variability due to
We extend the concept of CDF-series to the context of several variables, and show that the series solution of first-order differential equations y' = x(t,y) and functional equation y = x(t, y), with x CDF in two v...
详细信息
We extend the concept of CDF-series to the context of several variables, and show that the series solution of first-order differential equations y' = x(t,y) and functional equation y = x(t, y), with x CDF in two variables, are CDF-series. We also give many effective closure properties for CDF-series in several variables
Any device based on materials exhibiting circularly polarized luminescence (CPL) requires a high degree of circular polarization, quantified through the dissymmetry factor g(lum) and related quantities such as the CPL...
详细信息
Any device based on materials exhibiting circularly polarized luminescence (CPL) requires a high degree of circular polarization, quantified through the dissymmetry factor g(lum) and related quantities such as the CPL brightness. However, maximizing |g(lum)| is not an easy task. For a given chiral scaffold, for example, helicenes, g(lum) is usually confined below a certain threshold (e.g. 10(-3)-10(-2)). computational tools able to predict dissymmetry values in an accurate way are therefore of utmost importance. Intrigued by exceptionally high |g(lum)| values reported for two 1,1 & PRIME;-bitriphenylene-based [7]helicenes, we tested time-dependent density functional theory (TD-DFT) including solvation effects in predicting these large g(abs) and g(lum) values, in comparison with a reference [7]helicene. The quality of our results is discussed in comparison with a literature dataset.
The existence of at least four exact self-affine time functions, called E5:3 functions, that allow for an infinite number of exact replications of Growth-Decay 12345abc structures, is demonstrated. These E5:3 function...
详细信息
The existence of at least four exact self-affine time functions, called E5:3 functions, that allow for an infinite number of exact replications of Growth-Decay 12345abc structures, is demonstrated. These E5:3 functions are defined by algorithms and have no derivative anywhere. One of these E5:3 functions, called the standard E5:3 function, has the property of scaling like a random walk. This function also depends on the Golden Mean and rotates congruently with the Golden Spiral. A complete L-system specification for generating the standard E5:3 function is presented.
The primary goal of this paper is to present a new approximation theory. In classical rough sets, topological concepts, closure, and interior are expressible by elementary sets;hence, they are the elementary knowledge...
详细信息
The primary goal of this paper is to present a new approximation theory. In classical rough sets, topological concepts, closure, and interior are expressible by elementary sets;hence, they are the elementary knowledge approximations. However, for general binary relations, the corresponding closure and interior cannot always be interpreted by elementary knowledge. The primary results of this paper are to show that the appropriate generalized closure and interior are central knowledge approximations and have many expected nice properties, such as the upper approximation contains lower approximations. Here the central knowledge approximations mean the approximations are expressible by centers sets and the center set denotes the set of points that regard a binary neighborhood of the point p (which is the set of points that are (right) related to p) as its neighborhood. Many examples are used to justify our new view.
Various feature descriptions are being employed in logic programming languages and constraint-based grammar formalisms. The common notational primitive of these descriptions are functional attributes called features. ...
详细信息
Various feature descriptions are being employed in logic programming languages and constraint-based grammar formalisms. The common notational primitive of these descriptions are functional attributes called features. The descriptions considered in this paper are the possibly quantified first-order formulae obtained from a signature of binary and unary predicates called features and sorts, respectively. We establish a first-order theory FT by means of three axiom schemes, show its completeness, and construct three elementarily equivalent models. One of the models consists of the so-called feature graphs, a data structure common in computational linguistics. The other two models consist of the so-called feature trees, a record-like data structure generalizing the trees corresponding to first-order terms. Our completeness proof exhibits a terminating simplification system deciding validity and satisfiability of possibly quantified feature descriptions.
暂无评论