cheng@ral.rpi.edu (Wei-Ying Cheng) (01/04/91)
I would like to know if there is any reference about the conjugated gradient neural networks. If anyone knows, please post it in the news groups or tell me by E-mail. My E-mail address is: cheng@earth.ral.rpi.edu Thanks.
tylerh@nntp-server.caltech.edu (Tyler R. Holcomb) (01/04/91)
cheng@ral.rpi.edu (Wei-Ying Cheng) writes: >I would like to know if there is any reference about the conjugated >gradient neural networks. If anyone knows, please post it in the news >groups or tell me by E-mail. My E-mail address is: >cheng@earth.ral.rpi.edu >Thanks. "Improvement of the Backpropagation Algorithm for training Neural Networks", Leonard and Kramer, Computers and Chemical Engineering, Vol 14, No 3 pp 337-341, 1990
black@beno.CSS.GOV (Mike Black) (01/05/91)
There is at least one reference about a Scaled Conjugate Gradient method at cheops.cis.ohio-state.edu in pub/neurprose/moller.conjugate-gradient.ps.Z. There are quite a few good papers there. Here's a list of what they currently have: pollack.newraam.ps.Z pollack@cis.ohio-state.edu Preprint about the use of backprop auto-association recursively to generate patterns for symbolic trees. pollack.perceptrons.ps pollack@cis.ohio-state.edu Reprint of a somewhat negative review of Perceptrons, 1988 which appeared in J. Math Psych. pollack.nips88.ps pollack@cis.ohio-state.edu Reprint of NIPS 1988 paper suggesting how connectionism might be able to exploit chaos to overcome its symbolic limitations frean.upstart.ps marcus@cns.edinburgh.ac.uk The Upstart algorithm : a method for constructing and training feed-forward neural networks. jagota.hsn.ps.Z jagota@cs.buffalo.edu A Hopfield-style Network For Content-Addressable Memories, [Exponential Capacity, Graph Theory] witbrock.gf11sim.ps.Z mjw@cs.cmu.edu marcoz@cs.cmu.edu Implementation details and performance modelling for a 1.2 billion connections per second backprop simulator running on IBMs GF11 SIMD machine. (requires large memory postscript engine!) miikkulainen.discern.ps.Z risto@CS.UCLA.EDU A Neural Network Model of Script Processing and Memory miikkulainen.lexicon.ps.Z risto@CS.UCLA.EDU A Distributed Feature Map Model of the Lexicon jacobs.modular.ps.Z jacobs@gluttony.cs.umass.edu yu.output-biased.ps.Z yu@cs.utexas.edu Reprint of IJCNN90 paper showing how a faster and better learning can be achieved through specifying extra inputs. yu.epsilon.ps.Z yu@cs.utexas.edu Reprint of IJCNN90 paper presenting a way to decrease the errors in BP gradually to achieve a better correctness ratio. feldman.tr90-9.ps.Z jfeldman@icsi.berkeley.edu Proposal and discussion of an integrated language-vision learning task as a touchstone problem for Cognitive Science stolcke.tr90-15.ps.Z stolcke@icsi.berkeley.edu Investigation of simple recurrent networks as transducers that translate sequential NL input into semantic feature vectors gasser.morpho.ps.Z gasser@cs.indiana.edu This paper describes an extended version of a simple recurrent network which learns a linguistic rule, overcoming problems with previous connectionist approaches by incorporating both form and meaning in input and output. gasser.phonology.ps.Z gasser@cs.indiana.edu This paper describes the use of simple recurrent networks to learn phonological regularities in segmental linguistic input. stjohn.story.ps.Z stjohn%cogsci@ucsd.edu A recurrent network is used to model pronoun resolution and inference in text comprehension. sontag.capabilities.ps.Z sontag@hilbert.rutgers.edu This note deals with recognition capabilities of various feedforward architectures, analyzing the effect of direct input to output connections and comparing Heaviside and sigmoidal response units. ahmad.tr90-11.ps.Z ahmad@icsi.berkeley.edu "A Network for Extracting the Locations of Point Clusters Using Selective Attention". mani.function-space.ps.Z ganesh@cs.wisc.edu Learning by Gradient Descent in Function Space dietterich.comparison.ps.Z tgd@turing.cs.orst.edu A Comparison of ID3 and Backpropagation for English Text-to-Speech Mapping hampshire.bayes90.ps.Z John.Hampshire@SPEECH2.CS.CMU.EDU Equivalence proofs for MLP classifiers and the Bayesian Discriminant Fuction. jagota.tr90-25.ps.Z jagota@cs.buffalo.edu Hopfield-style network is characterised as a Maximal Cliques Graph Machine. Boolean Logic, Finite regular languages and other problems are mapped to it via transformations to graphs. cliff.manifesto.ps.Z Davec@cogs.sussex.ac.uk Arguments for computational neuroethology bengio.learn.ps.Z yoshua@cs.mcgill.ca/bengio@iro.umontreal.ca A method is presented for searching for and tuning, with learning methods, a synaptic learning rule which is biologically plausible, and yields networks capable to learn to perform difficult tasks. chalmers.evolution.ps.Z dave@cogsci.indiana.edu A Genetic algorithm rediscovers the delta rule. moller.conjugate-gradient.ps.Z fodslett@daimi.aau.dk A supervised learning algorithm (Scaled Conjugate Gradient, SCG) with superlinear convergence rate is introduced. SCG is fully automated including no user dependent parameters and avoids a time consuming line search, which other conjugate gradient algorithms use in order to determine a good step size. kruschke.gain.ps.Z kruschke@ucs.indiana.edu A paper to appear in Systems Man and Cybernetics pearlmutter.dynets.ps.Z barak.pearlmutter@cs.cmu.edu Survey of learning algorithms for recurrent neural networks with hidden units, with some new results and simulations. -- ------------------------------------------------------------------------------- : usenet: black@beno.CSS.GOV : land line: 407-494-5853 : I want a computer: : real home: Melbourne, FL : home line: 407-242-8619 : that does it all!: -------------------------------------------------------------------------------
sietsma@latcs1.oz.au (Jocelyn Sietsma Penington) (01/09/91)
In article <7#V^&F@rpi.edu> cheng@ral.rpi.edu (Wei-Ying Cheng) writes: >I would like to know if there is any reference about the conjugated >gradient neural networks. If anyone knows, please post it in the news The main reference I have used is A.R. Webb, D. Lowe and M.D. Bedworth, `A Comparison of nonlinear Optimization Strategies for Feed-Forward Adaptive Layered Networks' RSRE Memorandum 4157, Royal Signals and Radar Establishment, U.K., 1988. I'm afraid that's probably not very useful, because as you can see it is an internal report and has to be requested from R.S.R.E. There was a paper or two on it in proceedings of ICNN 1988, but I don't have the references to hand. Jocelyn -- (Jocelyn Penington, a.k.a. Sietsma - feel free to use either) Email: sietsma@LATCS1.oz.au Address: Materials Research Laboratory Phone: (03) 319 3775 or (03) 479 1057 PO Box 50, Melbourne 3032 This article does not commit me, LaTrobe Uni or M.R.L. to any act or opinion.