141.
A note on the lifting of linear and locally convex topologies on a quotient space
- Dierolf, Susanne
142.
Fitness-functions of Genetic Algorithms for optimizing Neural Network topologies
- Lars Kai Hansen
This paper presents a Genetic Algorithm (GA)
designed for finding topologies of recurrent networks
in order to optimize the performance in
terms of learning error and generalization ability.
In this context, the fitness of a chromosome
(network) is a function of its estimated test error,
i.e., its estimated generalization ability. For
this spesific GA, two different fitness-functions
are tested, one based on a validation set, and
one based on Akaikes's Final Prediction Error
estimate (FPE).
In this study, the task of the Neural Net is to
predict the sunspot activity based on the activity
of preceeding years. We find that the first
fitness-function does not give satisfactory evolution
for this task. The FPE based gives...
143.
Tree-Based Multicasting in Wormhole-Routed Irregular Topologies
- Ran Libeskind-hadas,Dominic Mazzoni,Ranjith Rajagopalan
A deadlock-free tree-based multicast routing algorithm
is presented for all direct networks, regardless of interconnection
topology. The algorithm delivers a message
to any number of destinations using only a single
startup phase. In contrast to existing tree-based
schemes, this algorithm applies to all interconnection
topologies, requires only fixed-sized input buffers
that are independent of maximum message length, and
uses a simple asynchronous flit replication mechanism.
The theoretical basis of the technique used here is sufficiently
general to develop other tree-based multicasting
algorithms for regular and irregular topologies. Simulation
results demonstrate that this tree-based algorithm
provides a very promising means of achieving
very low latency multicast.
1 Introduction
Recently, networks of workstations (NOWs) have
emerged as an inexpensive...
144.
Massively Parallel Training Of Multi Layer Perceptrons With Irregular Topologies
- D. Koll,M. Riedmiller,H. Braun
In this paper we present an approach to
the training of feed forward neural networks on massively
parallel SIMD-architectures. In order to cover
a wide field of applications we focus our attention
on the flexibility of the load balancing routines. Our
approach is characterized by three important properties:
1. All four types of parallelism inherent in
the training phase are used. 2. In a preprocessing
step neural networks are transformed into equivalent
topologies, more suited for parallel computation. 3.
Each learning task can be parallelized in a number of
different ways, the best of which is chosen according
to estimations of the computing efficiency.
Following these concepts we developed PINK
2
, a
massively parallel simulator...
145.
Fitness-functions of Genetic Algorithms for optimizing Neural Network topologies
- Lars Kai Hansen
This paper presents a Genetic Algorithm (GA)
designed for finding topologies of recurrent networks
in order to optimize the performance in
terms of learning error and generalization ability.
In this context, the fitness of a chromosome
(network) is a function of its estimated test error,
i.e., its estimated generalization ability. For
this spesific GA, two different fitness-functions
are tested, one based on a validation set, and
one based on Akaikes's Final Prediction Error
estimate (FPE).
In this study, the task of the Neural Net is to
predict the sunspot activity based on the activity
of preceeding years. We find that the first
fitness-function does not give satisfactory evolution
for this task. The FPE based gives...
146.
Interconnection Topologies and Routing for Parallel Processing Systems
- Gabriele Kotsis,Universit At Wien,Bitnet Adau Awiuni
The major aims of this work is to give a comparative survey of static interconnection
topologies, and to discuss their properties with respect to their use as interconnection topologies
in message passing multi-computer systems, i. e. each processing element has its own
local memory, there is no common memory, and the processing elements communicate via
message-passing. To this end it was necessary to recall relevant measures on graphs from
graph theory, like for example the average distance or the network diameter, and requirements
from the parallel processing area, like the reliability or extensibility. Special emphasis
has been given to present the construction rules for various graphs, because...
147.
Tree-Based Multicasting in Wormhole-Routed Irregular Topologies
- Ran Libeskind-hadas,Dominic Mazzoni,Ranjith Rajagopalan
Adeadlock-freetree-based multicast routing algorithm
is presented for all direct networks, regardless of interconnection
topology. The algorithm delivers a message
to any number of destinations using only a single
startup phase. In contrast to existing tree-based
schemes, this algorithm applies to all interconnection
topologies, requires only #xed-sized input bu#ers
that are independent of maximum message length, and
uses a simple asynchronous #it replication mechanism.
The theoretical basis of the technique used here is su#-
ciently general to develop other tree-based multicasting
algorithms for regular and irregular topologies. Simulation
results demonstrate that this tree-based algorithm
provides a very promising means of achieving
very low latency multicast.
1 Introduction
Recently, networks of workstations #NOWs# have
emerged as an inexpensive alternative...
148.
Tree-Based Multicasting in Wormhole-Routed Irregular Topologies
- Ran Libeskind-hadas,Dominic Mazzoni,Ranjith Rajagopalan
A deadlock-free tree-based multicast routing algorithm
is presented for all direct networks, regardless of interconnection
topology. The algorithm delivers a message
to any number of destinations using only a single
startup phase. In contrast to existing tree-based
schemes, this algorithm applies to all interconnection
topologies, requires only fixed-sized input buffers
that are independent of maximum message length, and
uses a simple asynchronous flit replication mechanism.
The theoretical basis of the technique used here is sufficiently
general to develop other tree-based multicasting
algorithms for regular and irregular topologies. Simulation
results demonstrate that this tree-based algorithm
provides a very promising means of achieving
very low latency multicast.
1 Introduction
Recently, networks of workstations (NOWs) have
emerged as an inexpensive...
149.
Tree-Based Multicasting in Wormhole-Routed Irregular Topologies
- Ran Libeskind-hadas,Dominic Mazzoni,Ranjith Rajagopalan
A deadlock-free tree-based multicast routing algorithm is presented for all direct
networks, regardless of interconnection topology. The algorithm delivers a message to
any number of destinations using only a single startup phase. In contrast to existing
tree-based schemes, this algorithm applies to all interconnection topologies, requires
only fixed-sized input buffers that are independent of maximum message length, and
uses a simple asynchronous flit replication mechanism. The theoretical basis of the
technique used here is sufficiently general to develop other tree-based multicasting
algorithms for regular and irregular topologies. Simulation results demonstrate that
this tree-based algorithm provides a very promising means of achieving very low latency
multicast.
1 Introduction
Recently, networks of workstations...
150.
Survivable routing of logical topologies in WDM networks
- Eytan Modiano,Aradhana Narula-tam
Network restoration is often done at the
electronic layer by rerouting traffic along a redundant path.
With Wavelength Division Multiplexing (WDM) as the
underlying physical layer, it is possible that both the primary
and backup paths traverse the same physical links and would
fail simultaneously in the event of a link failure. It is
therefore critical that lightpaths are routed in such a way that
a single link failure would not disconnect the network. We
call such a routing survivable and develop algorithms for
survivable routing of a logical topology. We prove necessary
and sufficient conditions for a routing to be survivable and
use this condition to formulate the problem as...
151.
Survivable routing of logical topologies in WDM networks
- Eytan Modiano,Aradhana Narula-tam
Network restoration is often done at the
electronic layer by rerouting traffic along a redundant path.
With Wavelength Division Multiplexing (WDM) as the
underlying physical layer, it is possible that both the primary
and backup paths traverse the same physical links and would
fail simultaneously in the event of a link failure. It is
therefore critical that lightpaths are routed in such a way that
a single link failure would not disconnect the network. We
call such a routing survivable and develop algorithms for
survivable routing of a logical topology. We prove necessary
and sufficient conditions for a routing to be survivable and
use this condition to formulate the problem as...
152.
Forming Connected Topologies in Bluetooth Ad-hoc Networks - An
- R. Guerin,J. Rank,S. Sarkar,E. Vergetis
This paper represents a first step in exploring the formation of connected topologies
in ad-hoc networks built on the Bluetooth technology. Connectivity is the most basic
requirement for any system aimed at allowing devices to communicate with each other and
in this paper we illustrate that this seemingly innocuous goal gives rise to many significant
challenges in the context of the Bluetooth technology. We start with a brief overview of
Bluetooth and its operation and then identify some of the major problems the technology
faces when used to build ad-hoc networks. The paper's contributions are in introducing
basic algorithmic problems associated with building connected Bluetooth networks and...
153.
Survivable routing of logical topologies in WDM networks
- Eytan Modiano,Aradhana Narula-tam
Network restoration is often done at the
electronic layer by rerouting traffic along a redundant path.
With Wavelength Division Multiplexing (WDM) as the
underlying physical layer, it is possible that both the primary
and backup paths traverse the same physical links and would
fail simultaneously in the event of a link failure. It is
therefore critical that lightpaths are routed in such a way that
a single link failure would not disconnect the network. We
call such a routing survivable and develop algorithms for
survivable routing of a logical topology. We prove necessary
and sufficient conditions for a routing to be survivable and
use this condition to formulate the problem as...
154.
oEvolve: Towards Evolutionary Overlay Topologies for High Bandwidth Data Dissemination
- Ying Zhu,Jiang Guo,Baochun Li
In this paper, we consider the problem of data dissemination
from a source to multiple receivers over applicationlayer
overlay networks, and seek to significantly improve end-toend
throughput of data dissemination sessions by constructing
topologies of high quality. We propose oEvolve, a distributed
algorithm that uses the strategy of progressively and adaptively
evolving the overlay topology over time towards high-quality
topologies, especially with respect to end-to-end throughput of
data dissemination. To validate the effectiveness and efficiency
of oEvolve, we present a fully distributed real-world oEvolve
implementation over PlanetLab, a global-scale wide-area overlay
network testbed. Our implementation consists of a framework of
components that involves a high-performance data forwarding
engine and a centralized performance monitoring...
155.
Searching for Edge Disjoint Paths on Hypercube-like Topologies
- Petr Kolman
Contents
1 Introduction 1
2 Preliminaries 5
2.1 Topologies and Embeddings : : : : : : : : : : : : : : : : : : : 5
2.2 Nonblocking Networks : : : : : : : : : : : : : : : : : : : : : : 10
2.3 Disjoint Paths Problem and On-line Algorithms : : : : : : : : 14
2.4 General Technique : : : : : : : : : : : : : : : : : : : : : : : : 17
3 Bene# Network 19
3.1 Searching for...
156.
An Adaptive Routing Algorithm for WK-Recursive Topologies
- Lorenzo Verdoscia,Roberto Vaccaro
This paper presents an easy and straightforward routing algorithm for WK-Recursive topologies. The
algorithm, based on adaptive routing, takes advantage of the geometric properties of such topologies.
Once a source node S and destination node D have been determined for a message communication, they
characterize, at some level l, two virtual nodes hl vn(SD ) and hl vn(D S ) that respectively contain S
but not D and D but not S. Such virtual nodes characterize other N d Gamma 2 (where N d is the node degree for
a fixed topology) virtual nodes hl vn(I SD ) of the same level that contain neither...
157.
A Simulator for SMT Architectures: Evaluating Instruction Cache Topologies
- Ronaldo Gonalves,Eduard Ayguad,Mateo Valero
SMT (Simultaneous MultiThreaded) is becoming one of
the major trends in the design of future generations of
microarchitectures. Its ability to exploit both intra- and interthread
parallelism makes it possible to exploit the potential
ILP (Instruction-level parallelism) that will be offered by
future processor designs. SMT architectures can hide high
latencies of instructions taking better advantage of the
hardware resources through the simultaneous execution of a
lot of diversified instructions from different threads. In order
to provide detailed and accurate information about the
performance of this approach, a SMT simulator has been
developed on top of the SimpleScalar Tool Set.
The SMT simulator allows the configuration of a large set
of architectural parameters...
158.
Sampling Internet Topologies: How Small Can We Go?
- Vaishnavi Krishnamurthy,Junhong Sun,Michalis Faloutsos,Sudhir Tauro
In this paper, we develop methods to "sample " a large real network into a small realistic graph. Although topology modeling has received a lot attention lately, it has not yet been completely resolved. Several methods create arguably realistic topologies from scratch. Our approach moves in the exact opposite direction. First, we observe that many real topologies are available to the networking community. However, their size makes them expensive to use in simulations as is. This brings up the following question: how can we shrink a graph, so that it still retains its essential properties? We propose an iterative sampling...
159.
On the Design of Fault-Tolerant Logical Topologies
- A. Nucci,B. Sanso,T. G. Crainic,E. Leonardi,M. Ajmone Marsan
In this paper we present a new methodology for
the design of fault-tolerant logical topologies in wavelengthrouted
optical networks supporting IP datagram flows. Our
design approach generalizes the "design protection" concepts,
and relies on the dynamic capabilities of IP to reroute
datagrams when faults occur, thus achieving protection
and restoration, and leading to high-performance costeffective
fault-tolerant logical topologies. In this paper for
the first time we consider resilience properties during the
logical topology optimization process, thus extending the optimization
of the network resilience also to the space of logical
topologies. Numerical results clearly show that our approach
outperforms previous ones, being able to obtain very
effective survivable logical topologies with limited computational
complexity.
160.
Small World and Scale-Free Network Topologies in an Artificial Regulatory Network
- P. Dwight Kuo,Wolfgang Banzhaf
Small world and scale--free network topologies commonly
exist in natural and artificial systems. Many mechanisms for
producing these topologies have been presented in the literature.
