Dynamic digital photorefractive memory for optoelectronic neural network learning modules

Neural network modules based on page-oriented dynamic digital photorefractive memory are described. The modules can implement two different interconnection organizations, fan-out and fan-in, depending on their target network applications. Neural network learning is realized by the real-time memory update of dynamic digital photorefractive memory. Physical separation of subvolumes in the page-oriented photorefractive memory architecture contributes to the low cross talk and high diffraction efficiency of the stored interconnection weights. Digitally encoded interconnection weights ensure high accuracy, providing superior neural network system scalability. Module scalability and feedforward throughput have been investigated based on photorefractive memory geometry and the photodetector power requirements. The following four approaches to extend module scalability are discussed: partial optical summation, semiparallel feedforward operation, time partitioning, and interconnection matrix partitioning. Learning capabilities of the system are investigated in terms of required interconnection primitives for implementing learning processes and three memory-update schemes. The experimental results of Perceptron learning network implementation with 900 input neurons with digital 6-bit accuracy are reported.     

Fast kernel conditional density estimation: A dual-tree Monte Carlo approach

We describe a fast, data-driven bandwidth selection procedure for kernel conditional density estimation (KCDE). Specifically, we give a Monte Carlo dual-tree algorithm for efficient, error-controlled approximation of a cross-validated likelihood objective. While exact evaluation of this objective has an unscalable O(n²) computational cost, our method is practical and shows speedup factors as high as 286,000 when applied to real multivariate datasets containing up to one million points. In absolute terms, computation times are reduced from months to minutes. This enables applications at much greater scale than previously possible. The core idea in our method is to first derive a standard deterministic dual-tree approximation, whose loose deterministic bounds we then replace with tight, probabilistic Monte Carlo bounds. The resulting Monte Carlo dual-tree algorithm exhibits strong error control and high speedup across a broad range of datasets several orders of magnitude greater in size than those reported in previous work. The cost of this high acceleration is the loss of the formal error guarantee of the deterministic dual-tree framework; however, our experiments show that error is still amply controlled by our Monte Carlo algorithm, and the many-order-of-magnitude speedups are worth this sacrifice in the large-data case, where cross-validated bandwidth selection for KCDE would otherwise be impractical.     

On properties not inherited by monoids from their Schützenberger groups

We give an example of a monoid with finitely many left and right ideals, all of whose Schützenberger groups are presentable by finite complete rewriting systems, and so each have finite derivation type, but such that the monoid itself does not have finite derivation type, and therefore does not admit a presentation by a finite complete rewriting system. The example also serves as a counterexample to several other natural questions regarding complete rewriting systems and finite derivation type. Specifically it allows us to construct two finitely generated monoids M and N with isometric Cayley graphs, where N has finite derivation type (respectively, admits a presentation by a finite complete rewriting system) but M does not. This contrasts with the case of finitely generated groups for which finite derivation type is known to be a quasi-isometry invariant. The same example is also used to show that neither of these two properties is preserved under finite Green index extensions.     

STRINGVACUA. A Mathematica package for studying vacuum configurations in string phenomenology

We give a simple tutorial introduction to the Mathematica package STRINGVACUA, which is designed to find vacua of string-derived or inspired four-dimensional N=1 supergravities. The package uses powerful algebro-geometric methods, as implemented in the free computer algebra system Singular, but requires no knowledge of the mathematics upon which it is based. A series of easy-to-use Mathematica modules are provided which can be used both in string theory and in more general applications requiring fast polynomial computations. The use of these modules is illustrated throughout with simple examples.

Program summary

Program title: STRINGVACUACatalogue identifier: AEBZ_v1_0Program summary URL:http://cpc.cs.qub.ac.uk/summaries/AEBZ_v1_0.htmlProgram obtainable from: CPC Program Library, Queen's University, Belfast, N. IrelandLicensing provisions: GNU GPLNo. of lines in distributed program, including test data, etc.: 31 050No. of bytes in distributed program, including test data, etc.: 163 832Distribution format: tar.gzProgramming language: “Mathematica” syntaxComputer: Home and office spec desktop and laptop machines, networked or stand aloneOperating system: Windows XP (with Cygwin), Linux, Mac OS, running Mathematica V5 or aboveRAM: Varies greatly depending on calculation to be performedClassification: 11.1External routines: Linux: The program “Singular” is called from Mathematica. Windows: “Singular” is called within the Cygwin environment from Mathematica.Nature of problem: A central problem of string-phenomenology is to find stable vacua in the four-dimensional effective theories which result from compactification.Solution method: We present an algorithmic method, which uses techniques of algebraic geometry, to find all of the vacua of any given string-phenomenological system in a huge class.Running time: Varies greatly depending on calculation requested.     

An Architecture for Human‐Guided Autonomy: Team TROOPER at the DARPA Robotics Challenge Finals

Recent robotics efforts have automated simple, repetitive tasks to increase execution speed and lessen an operator's cognitive load, allowing them to focus on higher‐level objectives. However, an autonomous system will eventually encounter something unexpected, and if this exceeds the tolerance of automated solutions, there must be a way to fall back to teleoperation. Our solution is a largely autonomous system with the ability to determine when it is necessary to ask a human operator for guidance. We call this approach human‐guided autonomy. Our design emphasizes human‐on‐the‐loop control where an operator expresses a desired high‐level goal for which the reasoning component assembles an appropriate chain of subtasks. We introduce our work in the context of the DARPA Robotics Challenge (DRC) Finals. We describe the software architecture Team TROOPER developed and used to control an Atlas humanoid robot. We employ perception, planning, and control automation for execution of subtasks. If subtasks fail, or if changing environmental conditions invalidate the planned subtasks, the system automatically generates a new task chain. The operator is able to intervene at any stage of execution, to provide input and adjustment to any control layer, enabling operator involvement to increase as confidence in automation decreases. We present our performance at the DRC Finals and a discussion about lessons learned.     

A Faà di Bruno Hopf algebra for a group of Fliess operators with applications to feedback

A Faà di Bruno type Hopf algebra is developed for a group of integral operators known as Fliess operators, where operator composition is the group product. Such operators are normally written in terms of generating series over a noncommutative alphabet. Using a general series expansion for the antipode, an explicit formula for the generating series of the compositional inverse operator is derived. The result is applied to analytic nonlinear feedback systems to produce an explicit formula for the feedback product, that is, the generating series for the Fliess operator representation of the closed-loop system written in terms of the generating series of the Fliess operator component systems. This formula is employed to provide a proof that local convergence is preserved under feedback.     

Nonparametric supervised learning by linear interpolation with maximum entropy

Nonparametric neighborhood methods for learning entail estimation of class conditional probabilities based on relative frequencies of samples that are "near-neighbors" of a test point. We propose and explore the behavior of a learning algorithm that uses linear interpolation and the principle of maximum entropy (LIME). We consider some theoretical properties of the LIME algorithm: LIME weights have exponential form; the estimates are consistent; and the estimates are robust to additive noise. In relation to bias reduction, we show that near-neighbors contain a test point in their convex hull asymptotically. The common linear interpolation solution used for regression on grids or look-up-tables is shown to solve a related maximum entropy problem. LIME simulation results support use of the method, and performance on a pipeline integrity classification problem demonstrates that the proposed algorithm has practical value.     

TreePlus: interactive exploration of networks with enhanced tree layouts

Despite extensive research, it is still difficult to produce effective interactive layouts for large graphs. Dense layout and occlusion make food Webs, ontologies and social networks difficult to understand and interact with. We propose a new interactive visual analytics component called TreePlus that is based on a tree-style layout. TreePlus reveals the missing graph structure with visualization and interaction while maintaining good readability. To support exploration of the local structure of the graph and gathering of information from the extensive reading of labels, we use a guiding metaphor of "plant a seed and watch it grow." It allows users to start with a node and expand the graph as needed, which complements the classic overview techniques that can be effective at (but often limited to) revealing clusters. We describe our design goals, describe the interface and report on a controlled user study with 28 participants comparing TreePlus with a traditional graph interface for six tasks. In general, the advantage of TreePlus over the traditional interface increased as the density of the displayed data increased. Participants also reported higher levels of confidence in their answers with TreePlus and most of them preferred TreePlus     

Unsupervised multiresolution segmentation for images with low depthof field

Unsupervised segmentation of images with low depth of field (DOF) is highly useful in various applications. This paper describes a novel multiresolution image segmentation algorithm for low DOF images. The algorithm is designed to separate a sharply focused object-of-interest from other foreground or background objects. The algorithm is fully automatic in that all parameters are image independent. A multi-scale approach based on high frequency wavelet coefficients and their statistics is used to perform context-dependent classification of individual blocks of the image. Unlike other edge-based approaches, our algorithm does not rely on the process of connecting object boundaries. The algorithm has achieved high accuracy when tested on more than 100 low DOF images, many with inhomogeneous foreground or background distractions. Compared with he state of the art algorithms, this new algorithm provides better accuracy at higher speed     

Comparative study of the pharmacokinetics of amitriptyline oxide and trimipramine after single administration in healthy male probands and patients with renal failure

The pharmacokinetics of the antidepressants amitriptyline oxide and trimipramine and their major metabolites amitriptyline, nor-triptyline and desmethyltrimipramine, were studied in twelve healthy male subjects (aged from 22 to 62 years) and twelve patients (aged from 25 to 73 years) with severe renal impairment (glomerular filtration rate < 10 ml/min). Oral single doses of 60 mg amitriptyline oxide and 50 mg trimipramine, separated by a washout period, were administered to all study participants. Blood and urine samples were collected up to 120 hours after administration. For trimipramine and desmethyltrimipramine, a new HPLC method was developed. The "Fischer Somatic and Undesired Effects Check List" was used for the assessment of adverse events. The mean plasma half-life and AUC of amitriptyline oxide and its metabolites were significantly higher in patients than in healthy adults. For trimipramine the AUC was significantly higher in patients. The plasma half-life of trimipramine was longer in patients, but statistically not significant. The maximum plasma concentrations for both drugs and metabolites were at an average distinctly higher in patients. Clearance rate of amitroptylinoxide and trimipramine also differed between the two groups. Correlating with these results a high incidence and a longer persistence (in most cases > 12 hours) and more pronounced adverse effects were noted in the patient group, whereas in volunteers adverse events were only observed up to approximately eight hours.