C. Stan ll and D. L. Waltz. Toward memory-based reasoning. Communications of the ACM, 29:1213-1228, December 1986.  Home/Search   Document Not in Database   Summary   ACM   TOC   Related Articles   Check

....functions, nonparametric techniques, for instance nearest neighbor method, must be used for classifications [21, 25] The nearest neighbor method is one of the simplest methods conceptually, and is commonly cited as a basis of comparison with other methods. It is often used in case based reasoning [62]. This section is devoted to statistical concept learning methods because they have similarities to the FIL algorithms developed in this thesis First, Bayes Decision Theory and Naive Bayesian Classifiers will be explained. Then, nearest neighbor methods with some variants will be discussed. ....

....feature values [54, 55] the most common approach is to set them to the mean value of the values on corresponding feature. Stanfill and Waltz introduced the Value Difference Metric (VDM) to define the similarity for symbolic valued (nominal) features and empirically demonstrated its benefits [62]. The VDM computes a distance for each pair of the different values a symbolic feature can assume. It essentially compares the relative frequencies of each pair of symbolic values across all classes. Two feature values have a small distance if their relative frequencies are approximately equal for ....

G. Stanfill and D. Waltz, Toward Memory-Based Reasoning, Communications of the ACM 29:1213-1228, 1986.

.... decision trees [55] have been the subject of much recent experimental work, the nearest neighbor algorithms continues to stay as an accurate learning technique [64] The nearest neighbor learning algorithms have been shown to work as well as other machine learning methods despite their simplicity [16, 18, 68]. It seems that nearest neighbor methods will continue to be cited as a basis of comparison with other methods. The NN classification algorithm is based on the assumption that examples which are closer in the instance space are of the same class. That is, unclassified ones should belong to the ....

....unknown (missing) feature values [57, 58] the most common approach is to set them to the mean of the values on corresponding feature. Stanfill and Waltz introduced the Value Difference Metric (VDM) to define the similarity for discrete (nominal) features and empirically demonstrated its benefits [68]. The VDM computes a distance for each pair of the different values a nominal feature can assume. It essentially compares the relative frequencies of each pair of distinct values across all classes. Two feature values have a small distance if their relative frequencies are approximately equal for ....

G. Stanfill and D. Waltz, Toward Memory--Based Reasoning, Communications of the ACM 29:1213--1228, 1986.

....feature vector in fX; Y g = f(X 1 ; Y 1 ) X 2 ; Y 2 ) Xn ; Yn )g closest to Z. The nearest neighbor decision rule classi es the unknown pattern Z into class Y j . A key feature of this decision rule (also called lazy learning [2] instance based learning [3] and memory based reasoning [100]) is that it performs remarkably well considering that no explicit knowledge of the underlying distributions of the data is used. Consider for example the two class problem and denote the a priori probabilities of the two classes by P (C 1 ) and P (C 2 ) the a posteriori probabilities by P (C 1 ....

....the harmonic mean coecient. It is also closely related to the Bayesian distance (Devijver [32] and the quadratic mutual information (Toussaint [108] Incidentally, the Bayesian distance is called the cross category feature importance in the instance based learning literature (Stan ll and Waltz [100], Creecy et al. 22] Furthermore, it is identical to the asymptotic probability of correct classi cation of the 1 NN rule given by P c [1 NN] 1 P e [1 NN] The error probability P e [1 NN] also shares a property with Shannon s measure of equivocation. Both are special cases of the ....

[Article contains additional citation context not shown here]

C. Stan ll and D. L. Waltz. Toward memory-based reasoning. Communications of the ACM, 29:1213-1228, December 1986.

....) 1) In other words, a point d ffl DN is positively classified by h hCB;oei if and only if there is a stored positive exemplar d pos that is strictly more similar to d according to the similarity measure oe than any of the stored negative exemplars d neg . Like many other IBL algorithms (e.g. [12] [5] 1] the learner studied here uses a weighted similarity measure; here, this measure is simply a sum of the weights of the bits of the representation on which two descriptions agree: oe w (d 1 ; d 2 ) P N w i X w i Theta (1 Gamma j(d 1 ) i Gamma (d 2 ) i j) 2) If the weight ....

C. Stanfill and D. L. Waltz. Towards memory-based reasoning. Communications of the ACM, 29(12):1213--1228, 1986.

....machine learning. The distinguishing feature of IBL is the fact that no explicit abstractions are constructed on the basis of the training examples during the training phase. A selection of the training items themselves is used to classify new inputs. IBL shares with Memory Based Reasoning (MBR, [Stanfill and Waltz, 1986]) and Case Based Reasoning (CBR, Riesbeck and Schank, 1989] the hypothesis that much of intelligent be haviour is based on the immediate use of stored episodes of earlier experience rather than on the use of explicitly constructed abstractions extracted from this experience (e.g. in the form of ....

....the other way round) 4 Discussion 4.1 Related Work As mentioned earlier, Instance Based Learning is a form of case based reasoning: a set of exemplars (cases) and a similarity metric are used to make de cisions about unseen cases. Earlier work on the ap plication of Memory Based Reasoning ([Stanfill and Waltz, 1986; Stanfill, 1987] another form of case based reasoning) to the phonemisation problem using the NetTalk data (MBRTalk) showed a better per formance than NetTalk itself ( Sejnowski and Rosenberg, 1987] however at the cost of an expensive, domain dependent computational measure of ....

C. W. Stunfill and D. Waltz. Toward memorybased reasoning. Communications of the ACM, 29:12, 1213-1228, 1986.

....in Section 5. Finally, our IIS design has been informed by earlier work in artificial immune systems. These many influences are briefly reviewed below. Case based reasoning is a technique that adapts solutions to past problems to solve similar current problems [57] Memory based reasoning [66] and instance based learning [1] are related schemes that use the solution of the most similar previous problem. Systems using these approaches learn by remembering specific past events rather than creating rules or generalizations. Immune memory uses a form of instance based learning; the ....

C. Stanfill and D. Waltz, "Toward memory-based reasoning," Communications of the ACM, vol. 29, no. 12, pp. 1213--1228, 1986.

....which is the usual way they have been tested) 3.1 Data Sets We selected ten data sets (Table 2) that have only numeric or boolean features, no missing values, and at least four classes. We avoided data sets with symbolic features because they often require distinct weighting metrics (e.g. Stanfill and Waltz, 1986 ] which complicates isolating the effects of feature selection. Data sets with fewer than four classes do not greatly benefit from ECOCs. Because few publicly available classification problems match these constraints, our selection is somewhat complete. These data sets are available from the ....

C. Stanfill and D. Waltz. Toward memory-based reasoning. Communication of ACM, 29:1213--1229, 1986.

....descendant of associative memory ideas. Memory based reasoning holds considerable promise, both for cognitive modeling and for applications. In this model, rote memories of episodes play the central role, and schemas are viewed as epiphenomenal. This model is described in considerable detail in [351 and will not be explained here; however, as I have prepared this paper, it has served as the background against which .I have critically examined both connectionist and more traditional AI paradigms. 2 Connectionist and Heuristic Search Models For most of its history, the heuristic search, ....

....is closest to a given current event or situation, it can exhibit highly regular behavior. Such systems degrade gracefully. Unlike connectionist models, associative memory models can also tell when a new event does not correspond well to any previous event, they can know that they don t know I35] See also Grpssberg [7] In contrast, systems based on logic, unification and exact matching are inevitably brittle (i.e. situations even slightly outside the realm of those encoded in the rules fail completely, and the system exhibits discontinuous behavior) We see no way to repair this ....

Stanfill, C., and Waltz, D.L., "Toward Memory-Based Reasoning," to appear in Communications of the AC.I, December 1986.

No context found.

C. Stan ll and D. L. Waltz. Toward memory-based reasoning. Communications of the ACM, 29:1213-1228, December 1986.

No context found.

Stan#ll, C., & Waltz, D. #1986#. Toward memory-based reasoning. Communications of the ACM, 29 #12#, 1213#1228.

No context found.

Stanfill, C. & Waltz, D. (1986), Toward memorybased reasoning, Communications of the ACM, 29(12):1213--1228.

No context found.

Stanfill, C and Waltz, D, 1986, Toward memory-based reasoning. Communications of the ACM 29(12), 1213--1228.

No context found.

C. Stanfill and D. Waltz. Toward memory-based reasoning. Communications of the ACM, 29:1213--1228, December 1986.

No context found.

C. Stanll and D. Waltz. Toward memory-based reasoning. Communications of the ACM, 29:12131228, December 1986.

No context found.

C Stanfill and D L Waltz. Towards memory-based reasoning. Communications of the ACM, 29(12):1213--1228, 1986.

No context found.

C. Stanfill and D. Waltz, `Toward memory-based reasoning', Communications of the ACM, 1213--1228, (1986).

No context found.

Stanfill C, Waltz D (1986) Toward memory-based reasoning. CACM 29(12):1213--1228

No context found.

C. Stanfill and D. Waltz. Toward memory-based reasoning. Communications of the ACM, pages 1213--1228, 1986.

No context found.

C. Stanfill. Toward memory-based reasoning. Communications of the ACM, 29:1213--1228, 1986. 27

No context found.

C. Stanfill, and D. Waltz, "Toward Memory-Based Reasoning," Communications of the ACM 29(12), 1986, , pp. 1213-1228.

No context found.

C.Stanfill and D.Waltz. Toward memory-based reasoning. Communications of the ACM, 29:12131228, 1986.

No context found.

Stanfill, C. and Waltz, D. (1986). Toward memory-based reasoning. Communications of the ACM, 29, pages 1213-1228.

No context found.

Stanfill, C. and Waltz, D. Toward memory-based reasoning. Communications of the ACM, 29(12), 1213-1228.

No context found.

C. Stan ll and D. Waltz. Toward memory-based reasoning. Communications of the ACM, 29:1213-1228, 1986.

No context found.

Stanfill, C. and Waltz, D. (1986). Toward memory-based reasoning. Communications of the ACM, 29, 12131228.

First 50 documents  Next 50

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