Rudolph, G., A Location-Independent ASOCS Model. BYU Master's Thesis, January 1991. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
On Integrating Inductive Learning with Prior Knowledge and.. - Giraud-Carrier (1994) (Correct)
....to the newer instances (see [MAR94] for details) Complete minimization is not reasonable, but AA1 attempts partial minimization through pairwise comparison of the training instance and instances in the current instance set. The correctness of this aspect of the algorithm has been proved elsewhere [RUD91a]. The correctness of the actual learning algorithm has not. Section 3 fills this gap. The result of the above preprocessing phase is a delete list and an add list containing the instances to be removed and added, respectively, from the current instance set to keep it consistent and somewhat ....
Rudolph, G. (1991). A Location-Independent ASOCS Model. Master's Thesis, Brigham Young University, Department of Computer Science.
Analysis of the Convergence and Generalization of AA1 - Giraud-Carrier, Martinez (1993) (Correct)
....to the newer instances (see [6] for details) Complete minimization is not reasonable, but AA1 attempts partial minimization through pairwise comparison of the training instance and instances in the current instance set. The correctness of this aspect of the algorithm has been proved elsewhere [10]. The correctness of the actual learning algorithm has not. Section 3 fills this gap. The result of the above preprocessing phase is a delete list and an add list containing the instances to be removed and added, respectively, from the current instance set to keep it consistent and somewhat ....
Rudolph, G. A Location-Independent ASOCS Model. Master's Thesis, Brigham Young University, Department of Computer Science, 1991.
An Efficient Static Topology For Modeling Asocs - Rudolph, Martinez (1991) (1 citation) Self-citation (Rudolph) (Correct)
....the input in order to come up with a best guess answer, such as nearest match, stochastic selection, etc. 3.2. Learning Mode: Changing The Network Function The goal of learning is to store a consistent, efficient representation of instances (including generalization) which have been presented [7]. This includes changing the network function. In learning mode, both the input and the desired network output (together called the new instance) are broadcast to the nodes. As in execution mode, the operation of the network is parallel, asynchronous, and the nodes are independent. Each node ....
....its stored instance with the new instance, and may change its function based on the result. A node may change its instance, may delete itself from the network, or not change at all. Following are four limited examples of how learning takes place. For an in depth discussion of LIA learning see [7]. Assume the initial network state in figure 1. The addition of each of the following four instances in order illustrates how a network operates in learning mode: 1) A B C = Z (2) A = Z (3) A B = Z (4) AB CD = Z A B C = Z. Because their respective A s are opposite from the new ....
[Article contains additional citation context not shown here]
Rudolph, G., A Location-Independent ASOCS Model. BYU Master's Thesis, January 1991.
Location-Independent Neural Network Models - Rudolph Self-citation (Rudolph) (Correct)
....step toward making LIT general enough to support most ANNs, because an arbitrary number of layers can be supported. Chapter 5 presents the LIA model which, like all ASOCS models, uses an inherently dynamic topology. This chapter presents essentially a journal version of Rudolph s MS Thesis [Rud91b]. However, the paper also goes beyond the MS Thesis in two ways: 1) The paper places LIA within the LIT framework, which was only beginning to be developed as a consequence of the Thesis, and 2) The paper provides formal descriptions of basic ASOCS instances, operators, instance sets, and ....
....desired nodes. 1. Transform the ANN 2. Embed the LI nodes in a tree arbitrary number of hidden layers LI Nodes m m Figure 1. General LIT Transformation LITs have been developed for backpropagation [Rum86] and Adaptive SelfOrganizing Concurrent Systems Adaptive Algorithm 2 [Mar91b] [Rud91b]. Transformations for other important ANNs are also being developed. LITs can potentially support a broad set of ANNs, thus allowing one efficient implementation strategy to support dynamic variations of many ANNs. This paper presents the transformations for two networks, the competitive learning ....
[Article contains additional citation context not shown here]
G. Rudolph. A Location-Independent ASOCS Model. MS Thesis, Brigham Young University, 1991.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC