The Infona portal uses cookies, i.e. strings of text saved by a browser on the user's device. The portal can access those files and use them to remember the user's data, such as their chosen settings (screen view, interface language, etc.), or their login data. By using the Infona portal the user accepts automatic saving and using this information for portal operation purposes. More information on the subject can be found in the Privacy Policy and Terms of Service. By closing this window the user confirms that they have read the information on cookie usage, and they accept the privacy policy and the way cookies are used by the portal. You can change the cookie settings in your browser.
One major goal of research on massively parallel networks of neuron-like processing elements is to discover efficient methods for recognizing patterns. Another goal is to discover general learning procedures that allow networks to construct the internal representations that are required for complex tasks. This paper describes a recently developed procedure that can learn to perform a recognition task...
In this paper we present a formal model of the Boltzmann machine and a discussion of two different applications of the model, viz. (i) solving combinatorial optimization problems and (ii) carrying out learning tasks. Numerical results of computer simulations are presented to demonstrate the characteristic features of the Boltzmann machine.
Using an informal presentation of the specification techniques being used in COBWEB-2's design, this paper has described the machine and the principles of its operation. An on-wafer communications network with a fault-tolerant packet routing mechanism has been described. These techniques. Cartesian routing with convex wrapping, are a substantial improvement on Catt's spiral (used in COBWEB-1)...
Deadlock and starvation are highly undesirable in packet switching networks. The communication processor presented in this paper was designed such that these phenomena can be proved not to occur. Deadlock is avoided using a new method called class climbing; fair usage of the classes and administration of the temporal order of arrival of the packets guarantee that no starvation can occur. The design...
This paper describes an architecture for rule-based interpretation of uncertain data, which is currently under development at our labs. Inference on uncertain input facts is a central topic in Al, with application, e.g., to the syntactic-semantic layers of speech understanding systems. The severe requirements of real-time applications dictate a parallel approach to this problem. The description covers...
This paper presents two extensions to the axlomatic architecture description language AADL [11], which together allow for a modular and concise specification of multiprocessor architectures at levels of abstraction ranging from compiler/operating-system interface down to register-transfer-level. The specification method is illustrated by an AADL-definition of a top-level view of DOOM, a distributed...
The data-flow model of computation is an attractive methodology for multiprocessor programming for it offers the potential for unlimited parallelism detection at no programmer's expense. It is here applied to a distributed architecture based on a commercially available microprocessor (the Inmos Transputer). In this project, we have integrated the high-level data driven principles of scheduling within...
Properties that make the binary cube an attractive interconnection network for many-processor machines include expansibility, a simple routing-algorithm, and a diameter that grows logarithmically with the number of processors. In this paper we introduce an operation called twist that diverts two edges. By applying a twist to a number of judiciously chosen edges a cube is transformed into a so-called...
We present a mathematically rigorous and, at the same time, convenient method for systolic design and derive alternative systolic designs for one expository matrix computation problem: matrix multiplication. Each design is synthesized from a simple program and a proposed layout of processors. The synthesis derives (1) a systolic parallel execution, (2) channel connections for the proposed processor...
A process is a pair 〈 A, X 〉 in which A is a set of symbols (the alphabet) and X is a non-empty prefix-closed subset of A* (the trace set). A process may be viewed as the specification of a mechanism: -symbols correspond to events that may occur. -traces correspond to sequences of events that may be observed when the mechanism is in operation. In this paper we show how...
Machines with distributed memory have the mapping problem — assigning processes to processors. In this paper we define the mapping problem as an optimization problem and discuss the question, how far is an optimum solution from an average or random solution. The term robustness is introduced and explained in detail with two examples, the SUPRENUM and the Hypercube architecture. For the SUPRENUM...
Large knowledge bases form an important applications area for parallel architectures. The special problems of this domain concern the movement and protection of large amounts of complex data in a hierarchy of storage devices and processing elements, in a manner which is sympathetic to the logical structure of the information. This structure is a reflection of the underlying knowledge representation...
We report experiments with a parallel architecture called ZAPP[2] simulated on several connected INMOS transputers. Besides the usual synthetic benchmarks (eg Nfib), our experiments covered a range of applications including matrix multiply, heuristic search and the 0–1 knapsack optimisation problem. Some of these applications cannot be adequately supported by the original ZAPP model. We report the...
We present a method for the synthesis of systolic arrays from a system of recurrent equations of a problem. The class of solved problems involves the inductive ones. For an inductive problem, the result sequence is calculated using its own elements : after its calculation, each element of this sequence is used as a data for other elements calculations. Therefore, the systolic arrays solutions are...
This paper explores methods for extracting parallelism from a wide variety of numerical applications. We investigate communications overheads and load-balancing for networks of transputers. After a discussion of some practical strategies for constructing occam programs, two case studies are analysed in detail.
We present a technique for synthesizing systolic arrays which have non-uniform data flow governed by control signals. The starting point for the synthesis is a Recurrence Equation with Linear Depencencies (RELD) which is a generalization of the simple recurrences encountered in mathematics. A large class of programs, including all (single and multiple) nested-loop programs can be described by such...
Set the date range to filter the displayed results. You can set a starting date, ending date or both. You can enter the dates manually or choose them from the calendar.