Network Generality, Training Required, and Precision Required

نویسندگان

  • John S. Denker
  • Ben S. Wittner
چکیده

We show how to estimate (1) the number of functions that can be implemented by a particular network architecture, (2) how much analog precision is needed in the connections in the network, and (3) the number of training examples the network must see before it can be expected to form reliable generalizations. Generality versus Training Data Required Consider the following objectives: First, the network should be very powerful and versatile, i.e., it should implement any function (truth table) you like, and secondly, it should learn easily, forming meaningful generalizations from a small number of training examples. Well, it is information-theoretically impossible to create such a network. We will present here a simplified argument; a more complete and sophisticated version can be found in Denker et al. (1987). It is customary to regard learning as a dynamical process: adjusting the weights (etc.) in a single network. In order to derive the results of this paper, however, we take a different viewpoint, which we call the ensemble viewpoint. Imagine making a very large number of replicas of the network. Each replica has the same architecture as the original, but the weights are set differently in each case. No further adjustment takes place; the "learning process" consists of winnowing the ensemble of replicas, searching for the one( s) that satisfy our requirements. Training proceeds as follows: We present each item in the training set to every network in the ensemble. That is, we use the abscissa of the training pattern as input to the network, and compare the ordinate of the training pattern to see if it agrees with the actual output of the network. For each network, we keep a score reflecting how many times (and how badly) it disagreed with a training item. Networks with the lowest score are the ones that agree best with the training data. If we had complete confidence in lCurrently at NYNEX Science and Technology, 500 Westchester Ave., White Plains, NY 10604 @) American Institute of Physics 1988

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

GDOP Classification and Approximation by Implementation of Time Delay Neural Network Method for Low-Cost GPS Receivers

Geometric Dilution of Precision (GDOP) is a coefficient for constellations of Global Positioning System (GPS) satellites. These satellites are organized geometrically. Traditionally, GPS GDOP computation is based on the inversion matrix with complicated measurement equations. A new strategy for calculation of GPS GDOP is construction of time series problem; it employs machine learning and artif...

متن کامل

The Relative generality and precision of Evidence Based Medical Infor-mation Resources in the Recovery of Diabetes Information

Background and Aim: Relative generality and precision are two important criteria for measuring the efficiency and performance of information retrieval systems. The aim of this study was to compare the integrity and location of evidence-based bases in the digital library of Hamedan University of Medical Sciences in data retrieval of diabetes.    Methods: The design of this research is cross-sect...

متن کامل

Experimental Study on the Precision Requirements of Rbf, Rprop and Bptt Training

Most neurocomputer architectures support only xed point arithmetic which allows a higher degree of VLSI integration but limits the range and precision of all variables. Up to now the eeect of this limitation on neural network training algorithms has been studied only for standard models like SOM or BP. This paper presents the results of an experimental study in which the precision requirements ...

متن کامل

Development of an in-cylinder processes model of a CVVT gasoline engine using artificial neural network

Today, employing model based design approach in powertrain development is being paid more attention. Precise, meanwhile fast to run models are required for applying model based techniques in powertrain control design and engine calibration. In this paper, an in-cylinder process model of a CVVT gasoline engine is developed to be employed in extended mean valve control oriented model and also mod...

متن کامل

Real-Time Kinematic Network of Tehran, from Design to Application

Following the request of the Tehran municipality and in order to provide the spatial information required in their various projects, a real-time kinematic network has been designed for Tehran. Based on the existing measures such as the dilution of precision at the network point positions, two different designs have been proposed. A minimum number of six GNSS stations are used in both of the pro...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 1987