U. M. Fayyad, N. Weir, and S. Djorgovski. SKICAT: A machine learning system for automated cataloging of large scale sky surveys. In Proceedings of the Tenth International Conference on Machine Learning (ICML-93), pages 112{ 119, 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....inputs to a manageable load for the more computationally intensive upper layers of the hierarchy which are responsible for such tasks as feature identi cation and nal classi cation. In practice, hierarchical classi cation systems have proven highly e ective in such tasks as stellar classi cation [7], planetary geology [8, 9] and medical image analysis [10] A number of recent projects in the computer security community have taken advantage of the power of hierarchical classi cation in the design of distributed intrusion detection systems [11, 12, 13, 14] The sensor we describe in this ....

U. M. Fayyad, N. Weir, and S. Djorgovski. SKICAT: A machine learning system for automated cataloging of large scale sky surveys. In Proceedings of the Tenth International Conference on Machine Learning (ICML-93), pages 112{ 119, 1993.

.... association rules falls within the purview of database mining [AIS93a] ABN92] HS94] MKKR92] S 93] Tsu90] also called knowledge discovery in databases [HCC92] Lub89] PS91b] Related, but not directly applicable, work includes the induction of classification rules [BFOS84] Cat91] FWD93] HCC92] Qui93] discovery of causal rules [CH92] Pea92] learning of logical definitions [MF92] Qui90] fitting of functions to data [LSBZ87] Sch90] and clustering [ANB92] C 88] Fis87] The closest work in the machine learning literature is the KID3 algorithm presented in [PS91a] If ....

Usama Fayyad, Nicholas Weir, and S.G. Djorgovski. Skicat: A machine learning system for automated cataloging of large scale sky surveys. In 10th Int'l Conf. on Machine Learning, June 1993.

....and often impossible due to the sheer volume of data. Maturing technologies in data mining, computer vision, and machine learning o er the potential to greatly enhance and, in many cases, enable new scienti c explorations of large image sets. Early JPL successes in this direction include SKICAT [1], a supervised classi cation tool for cataloging stars and galaxies in the POSS II sky survey, and JARtool [2] an adaptive recognition system that was applied to the problem of locating small volcanoes in the Magellan Venus dataset. It may come as a surprise, but in both of these large scale ....

U. Fayyad, N. Weir, and S. Djorgovski, \SKICAT: A machine learning system for the automated cataloging of large-scale sky surveys," in Tenth International Conference on Machine Learning, pp. 112-119, Morgan Kaufmann, 1993.

....attributes defined by astronomers are measured for each object in a set of training images. The tributes include intensity, area, stati moments, and so forth. data are constructed by asking astronomers to classify each of objects in the training set. Decision tree learning algorithms [5, 3, 9] are then applied learn a mapping from the measured attributes to the desired classification. For objects are too faint for astronomers to classify, the training data is obtained from higher resolution images or previous small scale surveys covering the corresponding portion of the sky. This ....

....in future research. 2 Imagery A fundamental objective of the was to provide mapping of surface of Venus. mapping was performed using synthetic aperture because of its ability to penetrate the dense cloud cover surrounding Venus. A complete description of the SAR imaging system is given in [5], so here will only summarize the most important characteristics: l Wavelength frequency: 2.385 S l Incidence Angle: 15 45 (nominal) l Range resolution: 120m 360m 3 l Azimuth resolution: 120m l Pixel spacing: 75m (full resolution l Number of looks: 5 16 Figure 1 shows a 30km x ....

Fayyad, N. Weirand, and S. Djorgovski. Skicat: A machine learning system for the automated cataloging of large-scale sky surveys, In Tenth International Conference on Machine Learning,

....nineteen data sets, as well as some others (Jensen 1998) a plateau was reached after very few training examples. Of course, when there exists a massive volume of data, some sampling may be necessary, whether or not it decreases accuracy. For example, in a famous application of inductive learning, Fayyad et al. 1993) used sampling techniques (among others) to reduce more than three terabytes of raw data. Therefore, it is important to consider whether it is possible to sample efficiently. Consider that if it is necessary to scan the entire data set in order to produce a random sample, much of the advantage of ....

....the data set randomly into n subsets. Figure 5 shows a general model of partitioned data learning. More precisely, this figure shows a model of independent multi subset learning, because there is no interaction between the n learning runs; the C i are formed independently, and then combined. Fayyad et al. 1993) use a sequential independent multi subset approach in which the L i are decision tree learners; the C i are rule sets extracted from the decision trees, and the combination procedure is a greedy covering algorithm. When multiple subsets are being processed sequentially, it is possible to take ....

[Article contains additional citation context not shown here]

Fayyad, U., N. Weir, and S. Djorgovski (1993). SKICAT: A machine learning system for automated cataloging of large scale sky surveys. In Proceedings of the Tenth International Conference on Machine Learning. Morgan Kaufmann.

.... features (e.g. crop types) in order to move from speci c examples to generalizations [45] Other algorithms apply machine learning techniques to automatically uncover such trends; for example, by noticing common features of example clusters labeled as being stars in a database of sky images [24]. Generalizations serve as the basis for inducing classi cation rules (e.g. decision trees [62] to apply to new data sets. These techniques have proved quite useful in analyzing massive scienti c data sets; for example, in automatically cataloging sky images [24] identifying volcanos in images ....

....stars in a database of sky images [24] Generalizations serve as the basis for inducing classi cation rules (e.g. decision trees [62] to apply to new data sets. These techniques have proved quite useful in analyzing massive scienti c data sets; for example, in automatically cataloging sky images [24], identifying volcanos in images of the surface of Venus [11] and tracking cyclones in weather data [75] It remains interesting future research to explore expressions of such generalization and classi cation algorithms in the Spatial Aggregation Language. 2.3 Spatial Reasoning Applications such ....

[Article contains additional citation context not shown here]

U. Fayyad, N. Weir, and S. Djorgovski. SKICAT: a machine learning system for the automated cataloging of large-scale sky surveys. In Proceedings of 10th International Conference on Machine Learning, 1993.

....of the data as well as patterns that may be exhibited in the data. The field of machine learning has made substantial progress over the years and a number of algorithms have been popularized and applied to a host of applications in diverse fields (Langley Simon, 1995; Bratko Muggleton, 1995; Fayyad et al. 1993; Craven Shavlik, 1994) Thus, we may simply apply the current generation of learning algorithms to very large databases and wait for a response However, the question is how long might we wait Indeed, do the current generation of machine learning algorithms scale from tasks common today that ....

Fayyad, U., Weir, N., & Djorgovski, S. (1993). SKICAT: A machine learning system for automated cataloging of large scale sky surveys. Proc. Tenth Intl. Conf.

.... problem has been mentioned by several other authors, including Burl et al. this issue) Cherkauer and Shavlik (1994) Ezawa et al. 1996) Fawcett and Provost (1997) Kubat, Pfurtscheller and Flotzinger (1994) and Pfurtscheller, Flotzinger and Kalcher (1992) For instance, in the SKICAT system (Fayyad, Weir Djorgovski, 1993), the batches were plates, from which image regions were selected. When the system trained on images from one plate was applied to images from another plate, the classification accuracy dropped well below that of manual classification. The solution used in SKICAT was to normalize some of the ....

Fayyad, U.M., Weir, N., & Djorgovski, S. (1993). SKICAT: A Machine Learning System for Automated Cataloging of Large Scale Sky Surveys. Proceedings of the Tenth International Conference on Machine Learning (pp. 112--119), Morgan Kaufmann.

....of the final tree, but not its accuracy. Variations in the splitting criterion: The traditional decision tree formulation splits on one attribute at a time and creates branches for all possible values of that attribute, but variations on this are certainly possible. Default branches Fayyad [32, 34] shows that there are a number of disadvantages to creating branches for all values of an attribute (such as the irrelevant values problem , the reduced training data problem , and the missing branches problem) One possibility is to branch on specific values of the attribute, and leave other ....

Usama M. Fayyad, Nicholas Weir, and S. Djorgovski. Skicat: A machine learning system for automated cataloging of large scale sky surveys. In Proceedings of the Tenth International Conference on Machine Learning, Amherst, MA, pages 112--119. Morgan Kaufmann, San Mateo, CA, 1993.

....concluded that they are not a solution to the general problem of scaling up to very large data sets. However, when there exists a massive volume of data, some sampling may be necessary. For example, in one of the more famous applications of inductive learning to massive data sets, Fayyad, et al. [31], used sampling techniques, among others, to reduce over three terabytes of raw data. Catlett [15] 16] also studied the use of sampling tactically to reduce complexity as learning algorithms process large data sets. In decision tree induction, searching for good split values for numeric ....

....in Section 6.3. Figure 5 showed a general model of partitioned data learning. More precisely, this figure showed a model of independent multi sample learning, because it shows no interaction between the n learning runs; the C i are formed independently, and then combined. Fayyad, et al. [31], use an independent multi sample approach in which the combination procedure is a covering algorithm. When multiple samples are being processed sequentially, it is possible for approaches to take advantage of knowledge learned in one iteration to guide learning in the next iteration. Figure 6 ....

[Article contains additional citation context not shown here]

Fayyad, U.M., Weir, N. and Djorgovski, S. (1993). SKICAT: A Machine Learning System for Automated Cataloging of Large Scale Sky Surveys. In Proc. of the Tenth Intl. Conf. on Machine Learning, Morgan Kaufmann.

....will fall hopelessly behind the number, variety, and complexity of the signal databases and the interests of the browsers. In the recent past there have been some attempts at automating the recognition and classification process. This automation has taken the form of machine learning (e.g. (Fayyad et al. 1993; Rao et al. 1993; Pazzani et al. 1994) and statistical techniques (e.g. Hunter, 1990; Agarwal et al. 1992) The goal is typically to acquire knowledge from large databases. The Skicat work is one of the few examples of machine learning applied to full resolution photo images (Fayyad et ....

....et al. 1993; Rao et al. 1993; Pazzani et al. 1994) and statistical techniques (e.g. Hunter, 1990; Agarwal et al. 1992) The goal is typically to acquire knowledge from large databases. The Skicat work is one of the few examples of machine learning applied to full resolution photo images (Fayyad et al. 1993). We know, however, of no published research that has accomplished the examination and retrieval of arbitrary database elements as large and unprocessed as real image and acoustic data. Any method for data mining has, at its heart, a function or algorithm for determining whether a particular ....

Fayyad, U. M., Weir, N., and Djorgovski, S. (1993). Skicat: A machine learning system for automated cataloging of large scale sky surveys. In Proceedings of the Tenth International Conference on Machine Learning. Morgan Kauffman.

.... association rules falls within the purview of database mining [AIS93a] ABN92] HS94] MKKR92] S 93] Tsu90] also called knowledge discovery in databases [HCC92] Lub89] PS91b] Related, but not directly applicable, work includes the induction of classification rules [BFOS84] Cat91] FWD93] HCC92] Qui93] discovery of causal rules [CH92] Pea92] learning of logical definitions [MF92] Qui90] fitting of functions to data [LSBZ87] Sch90] and clustering [ANB92] C 88] Fis87] The closest work in the machine learning literature is the KID3 algorithm presented in [PS91a] If ....

Usama Fayyad, Nicholas Weir, and S.G. Djorgovski. Skicat: A machine learning system for automated cataloging of large scale sky surveys. In 10th Int'l Conf. on Machine Learning, June 1993.

....and communicate to others, and they can also be used as the basis for an experience based decision support system. Algorithms for learning and refining classification rules from examples include the AQ family (Michalski et al. 1983, 1986) the ID3 family (Quinlan 1993, 1986; Cheng et al. 1988; Fayyad et al. 1993, 1994) and CN2 (Clark and Niblett 1989) The continuing development of rule induction algorithms is motivated by the increasing application of knowledge discovery from database methods (Fayyad et al. 1993; Gemello and Mana 1989; Piatetsky Shapiro et al. 1991) which apply inductiveinference ....

.... family (Michalski et al. 1983, 1986) the ID3 family (Quinlan 1993, 1986; Cheng et al. 1988; Fayyad et al. 1993, 1994) and CN2 (Clark and Niblett 1989) The continuing development of rule induction algorithms is motivated by the increasing application of knowledge discovery from database methods (Fayyad et al. 1993; Gemello and Mana 1989; Piatetsky Shapiro et al. 1991) which apply inductiveinference techniques to large databases. Sets of real world training examples are almost invariably imprecise. In such a situation, exact classification rules cannot be derived. The objective of this paper is to suggest a ....

Fayyad, U.M., Weir, N. and Djorgovski, S. 1993. "SKICAT: A Machine Learning System for Automated Cataloging of Large Scale Sky Surveys," Machine Learning: Proc. of the 10th International Conference.

....out of 20,000 examples) led 24 (4500 out of 30,000 examples) and census income (9768 out of 32561 examples) Of course, when there exists a massive volume of data, some sampling may be necessary, whether or not it decreases accuracy. For example, in a famous application of inductive learning, Fayyad et al. 1993) used sampling techniques (among others) to reduce more than three terabytes of raw data. Therefore, it is important to consider whether it is possible to sample efficiently. Consider that if it is necessary to scan the entire data set in order to produce a random sample, much of the advantage of ....

....the data set randomly into n subsets. Figure 5 shows a general model of partitioned data learning. More precisely, this figure shows a model of independent multi sample learning, because there is no interaction between the n learning runs; the C i are formed independently, and then combined. Fayyad et al. 1993) use a sequential independent multi sample approach in which the L i are decision tree learners; the C i are rule sets extracted from the decision trees, and the combination procedure is a greedy covering algorithm. When multiple samples are being processed sequentially, it is possible to take ....

[Article contains additional citation context not shown here]

Fayyad, U., N. Weir, and S. Djorgovski (1993). SKICAT: A machine learning system for automated cataloging of large scale sky surveys. In Proceedings of the Tenth International Conference on Machine Learning. Morgan Kaufmann.

....is to select a smaller sample from the initial data set. Catlett (1991b) studied a variety of strategies for sampling from a large data set. Despite the advantages of certain sampling strategies, Catlett concluded that they are not a solution to the problem of scaling up to very large data sets. Fayyad, et al. 1993), use sampling techniques, inter alia, to reduce a huge data set (over 3 terabytes of raw data) One method they use is to partition the data set, learn rules from subsamples, and use a covering algorithm to combine the rules. This method is similar to incremental batch learning and coarse grained ....

Fayyad, U., Weir, N., & Djorgovski, S. (1993). SKICAT: A Machine Learning System for Automated Cataloging of Large Scale Sky Surveys. In Proc. of the Tenth Int. Conf.

....or a high percentage of the total instances. A set of n classifiers are produced from n sets of samples and they are combined by voting or weighted voting. Ali and Pazzani (1996) use k fold partitioning to generate k models by training on all but the ith partition k times. In another approach, Fayyad, Weir and Djorgovski (1993) use a covering algorithm to combine the rulesets induced from several sets of random subsamples. The multiple models are usually produced from a single learning algorithm, though there is no such restriction in this formalism. Multiple models can also be produced by varying the learning ....

....and NB (Ting, 1994; 1996a) are used in our experiments. IB1 is a variant of IB1 (Aha, Kibler Albert, 1991) that incorporates the modified value difference metric (Cost Salzberg, 1993) and NB is an implementation of the Naive Bayes (Cestnik, 1990) algorithm. Both algorithms include a method (Fayyad Irani, 1993) for discretising continuous valued attributes in the preprocessing. This preprocessing improved the performance of the two algorithms in most of the continuous valued attribute domains studied by Ting (1994) We use the nearest neighbour for making prediction in IB1 and the default settings are ....

Fayyad, U.M., N. Weir & S. Djorgovski (1993), SKICAT: A Machine Learning System for Automated Cataloging of Large Scale Sky Surveys, in Proceedings of the Tenth International Conference on Machine Learning, pp. 112-119, Morgan Kaufmann.

....cannot be determined by visual inspection or classical computational approaches in astronomy. Our goal was to classify objects that are at least one isophotal magnitude fainter than objects classified in previous comparable surveys. We tackled the problem using decision tree learning algorithms [11] to accurately predict the classes of objects. Accuracy of the procedure was verified by using a very limited set of high resolution CCD images as ground truth. By extracting rules via statistical optimization over multiple trees [11] we were able to achieve 94 accuracy on predicting sky object ....

....We tackled the problem using decision tree learning algorithms [11] to accurately predict the classes of objects. Accuracy of the procedure was verified by using a very limited set of high resolution CCD images as ground truth. By extracting rules via statistical optimization over multiple trees [11] we were able to achieve 94 accuracy on predicting sky object classes. Reliable classification of faint objects increased the the size of data that is classified (usable for analysis) by 300 . Hence astronomers were able to extract much more out of the data in terms of new scientific results ....

U.M. Fayyad, N. Weir, and S. Djorgovski (1993) Skicat: A machine learning system for the automated cataloging of large-scale sky surveys. In Proc. of 10th International Conference on Machine Learning, pp 112-- 119.

No context found.

U. Fayyad, N. Weir, and S.G. Djorgovski (1993). "SKICAT: a machine learning system for automated cataloging of large scale sky surveys." Proc. of Tenth Int. Conf. on Machine Learning, Morgan Kaufman.

No context found.

U.M. Fayyad, N. Weir, and S. Djorgovski. SKICAT: A machine learning system for automated cataloging of large scale sky surveys. In Machine Learning: Proceedings of the Tenth International Conference, pages 112-- 119. Morgan Kaufmann, San Mateo, CA, 1993.

No context found.

Usama M. Fayyad, Nicholas Weir, and D. Djorgovski. SKICAT: A machine learning system for automated cataloging of large scale sky surveys. In ML-93 [330], pages 112--119. Editor: Paul E. Utgoff.

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