home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai.neural-nets
- Path: sparky!uunet!charon.amdahl.com!pacbell.com!decwrl!purdue!yuma!lamar!tstanley
- From: tstanley@lamar.ColoState.EDU (Thomas R Stanley)
- Subject: Help, statistics and learning algorithms
- Sender: news@yuma.ACNS.ColoState.EDU (News Account)
- Message-ID: <Jan26.205706.73285@yuma.ACNS.ColoState.EDU>
- Date: Tue, 26 Jan 1993 20:57:06 GMT
- Nntp-Posting-Host: lamar.acns.colostate.edu
- Organization: Colorado State University, Fort Collins, CO 80523
- Lines: 57
-
- Hi everybody:
-
- I am currently using classification procedures like Linear Discriminant
- Analysis (LDA) and Classification And Regression Trees (CART) as model
- selection procedures for a known (finite) set of multinomial models. In
- general, I generate data under a known model using Monte Carlo
- simulations, compute diagnostic statistics from the simulated data, then
- use these statistics as training data for LDA and CART. Thus, given a set
- of new data where the underlying model is unknown, I compute the
- diagnostic statistics and let LDA and CART tell me which model the new
- data were most likely to have come from. Lately, I have become interested
- in using neural nets for model selection under a supervised learning
- algorithm. My question is, what are the statistical equivalents of the
- supervised learning methods commonly used in constructing neural nets?
- More precisely, are there statistical equivalents to the following (see,
- I did check the FAQ first :) learning methods:
-
- 2. SUPERVISED LEARNING (i.e. with a "teacher"):
- 1). Feedback Nets:
- a). Brain-State-in-a-Box (BSB)
- b). Fuzzy Cognitive Map (FCM)
- c). Boltzmann Machine (BM)
- d). Mean Field Annealing (MFT)
- e). Recurrent Cascade Correlation (RCC)
- f). Learning Vector Quantization (LVQ)
- 2). Feedforward-only Nets:
- a). Perceptron
- b). Adaline, Madaline
- c). Backpropagation (BP)
- d). Cauchy Machine (CM)
- e). Adaptive Heuristic Critic (AHC)
- f). Time Delay Neural Network (TDNN)
- g). Associative Reward Penalty (ARP)
- h). Avalanche Matched Filter (AMF)
- i). Backpercolation (Perc)
- j). Artmap
- k). Adaptive Logic Network (ALN)
- l). Cascade Correlation (CasCor)
-
- I found one reference that said BP was equivalent to least squares
- fitting. Is this true of the rest of the methods? If these methods are
- really just variations on classical statistical procedures, then (and here
- is where I show my ignorance) what do I have to gain by using neural nets?
- Why should I expect these procedures to perform better than a parametric
- method (let's assume for the sake of argument the assumptions of the
- method (e.g. normality) are met) where there exists maximum likelihood
- estimators (MLE's) for the parameters (i.e. weights at the processing
- element or node)? MLE's guarantee unbiasedness and efficiency (minimum
- variance), you can't get better than that can you? I would appreciate any
- help, ideas, or input I can get on these questions. I don't want to spend
- a lot of time looking into neural nets if they are unlikely to lead to a
- better model selection procedure. Thanks in advance.
-
- Tom Stanley
-
- National Ecology Research Center
- Fort Collins, CO
-