| C. Cardie, \Using cognitive biases to guide feature set selection," in Proceedings of the Fourteenth Annual Conference of the Cognitive Science Society, pp. 469-471, Lawrence Erlbaum Associated, 1993. 18 |
.... choice of the inductive bias has two major drawbacks: 1) It is conceivable that di erent applications require di erent choices of the inductive bias H which leads to fast convergence and good generalization performance, and (2) it does not provide a Bias interactions have also been studied in [8, 9]. 9 mechanism for dealing with uncertainty about the initial domain theory. This section proposes a method for choosing the strength of the inductive bias which takes these two objections into account: The choice of H depends on the application represented by the initial domain theory, the ....
C. Cardie, \Using cognitive biases to guide feature set selection," in Proceedings of the Fourteenth Annual Conference of the Cognitive Science Society, pp. 469-471, Lawrence Erlbaum Associated, 1993. 18
....open, and appropriate representational biases and search heuristics that find sinple hypotheses are still required. Note that biases nay interact; a procedural and a representational bias nfight interact synergistically or conflict. Few researchers have studied bias interactions (see Cardie [8] and Cobb [11] for exceptions) Hopefully, future work will explore this inportant topic. Both representational and procedural biases can be evaluated (enpirically or analytically) by deternfining the effect they have or are expected to have on learning perfornance. Bias selection involves using ....
Cardie, C. (1993). Using cognitive biases to guide feature set selection. li Proceedings of the Fourteenth Annual Conference of the Cognitive Science Society, pages 743 748. Lawrence Erlbatm Associates.
....open, and appropriate representational biases and search heuristics that find simple hypotheses are still required. Note that biases may interact; a procedural and a representational bias might interact synergistically or conflict. Few researchers have studied bias interactions (see Cardie [8] and Cobb [11] for exceptions) Hopefully, future work will explore this important topic. Both representational and procedural biases can be evaluated (empirically or analytically) by determining the effect they have or are expected to have on learning performance. Bias selection involves using ....
Cardie, C. (1993). Using cognitive biases to guide feature set selection. In Proceedings of the Fourteenth Annual Conference of the Cognitive Science Society, pages 743--748. Lawrence Erlbaum Associates.
No context found.
Cardie, C. (1992c) "Using Cognitive Biases to Guide Feature Set Selection" Proceedings, Fourteenth Annual Conference of the Cognitive Science Society. pp. 743-748.
....terminal nodes, and bases most predictions on these stored cases. Because the system uses probabilistic weights to sort test cases, it effectively assigns different weights to each path in its tree and, like our approach, can apply a different similarity metric to each test case. In earlier work (Cardie 1992), we implemented a partially automated version of the cognitive bias approach for Cobweb and obtained similar results to those reported here. Global feature weighting methods that compute weights based on information gain have proved particularly helpful across a variety of natural language ....
C. Cardie. 1992. Using cognitive biases to guide feature set selection. In Proceedings of the Fourteenth Annual Conference of the Cognitive Science Society, pages 743--748, Indiana Univeristy, Bloomington, IN. Lawrence Erlbaum Associates. Also in Working Notes of the 1992 AAAI Workshop on Constraining Learning with Prior Knowledge, San Jose, CA.
No context found.
Cardie, C. (1992c). "Using Cognitive Biases to Guide Feature Set Selection" in the Proceedings of the AAAI92 Workshop on Constraining Learning with Prior Knowledge. San Jose, CA. pp. 11-18.
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