We propose a new network growing method to detect salient features in input patterns. The new method is based upon a previous network growing model (R. Kamimura and T. Kamimura, 2002) and introduced to overcome some p...
详细信息
We propose a new network growing method to detect salient features in input patterns. The new method is based upon a previous network growing model (R. Kamimura and T. Kamimura, 2002) and introduced to overcome some problems in the previous model. We have so far tried to build a model that can learn input patterns as efficiently as possible. To realize this efficiency, we impose upon networks a constraint that only connections into new competitive units must be updated to absorb as much information as possible from outside. However, one of the problems is that the previous improper feature extraction prevents networks from extracting appropriate features in the later learning stages. To overcome this problem, we relax the condition of the previous model, and we permit networks to update all connections for gradual feature extraction at the expense of computational efficiency. We applied the new method to a simple problem that the previous model cannot solve, and information education data analysis. In both problems, we found that the new method can appropriately extract features from input patterns.
In this paper, we propose a new information theoretic network growing algorithm. The new approach is called greedy information acquisition, because networks try to absorb as much information as possible in every stage...
详细信息
In this paper, we propose a new information theoretic network growing algorithm. The new approach is called greedy information acquisition, because networks try to absorb as much information as possible in every stage of learning. In the first stage, two competitive units compete with each other by maximizing mutual information. In the successive stages, new competitive units are gradually added and information is maximized. Through greedy information maximization, different sets of important features in input patterns can be cumulatively discovered in successive stages. We applied our approach to a language classification problem. Experimental results confirmed that different features in input patterns are gradually discovered.
作者：
Shneiderman, BenDepartment of Computer Science
Human-Computer Interaction Laboratory Institute for Advanced Computer Studies and Institute for Systems Research University of Maryland College ParkMD20742 United States
The growing use of information visualization tools and data mining algorithms stems from two separate lines of research. Information visualization researchers believe in the importance of giving users an overview and ...
详细信息
Users who must combine demographic, economic or other data in a geographic context are often hampered by the integration of tabular and map representations. Static, paper-based solutions limit the amount of data that ...
详细信息
Users who must combine demographic, economic or other data in a geographic context are often hampered by the integration of tabular and map representations. Static, paper-based solutions limit the amount of data that can be placed on a single map or table. By providing an effective user interface, we believe that researchers, journalists, teachers, and students can explore complex data sets more rapidly and effectively. This paper presents Dynamaps, a generalized map-based information visualization tool for dynamic queries and brushing on choropleth maps. Users can use color coding to show a variable on each geographic region, and then filter out areas that do not meet the desired criteria. In addition, a scatterplot view and a details-on-demand window support overviews and specific fact-finding.
In exploring how to make programming easier for non-programmers, research into end-user environments has traditionally been concerned with designing better human-computerinteraction. That traditional focus has left o...
详细信息
ISBN:
(纸本)0780371984
In exploring how to make programming easier for non-programmers, research into end-user environments has traditionally been concerned with designing better human-computerinteraction. That traditional focus has left open the question of how end-user environments might support human-humaninteraction. Especially in situations in which end-user environments are enlisted to facilitate learning, we hypothesize that a key benefit may be their ability to mediate conversations about a domain of interest. In what ways might end-user environments support human communication, and what design features make them well-suited to do so? Drawing on ethnographic studies of an undergraduate algorithms course in which students constructed and presented algorithm visualizations, we develop a provisional framework of six communicative dimensions of end-user environments: programming salience, typeset fidelity, story content, modifiability, controllability and referencability. To illustrate the design implications of these dimensions, we juxtapose conventional algorithm visualization technology with a prototype end-user environment specifically designed to facilitate communication about algorithms. By characterizing those aspects of end -user environments that impact social interaction, our framework provides an important extension to T.R.G. Green and M. Petre's (1996) cognitive dimensions.
A challenge for human-computerinteraction researchers and user interface designers is to construct information technologies that support creativity. This ambitious goal can be attained by building on an adequate unde...
详细信息
Traditional algorithm visualization software supports the creation of "high fidelity" visualizations, which depict the target algorithm for arbitrary input, and have the polished look of textbook figures. Dr...
详细信息
作者：
Ben ShneidermanDepartment of Computer Science
Human-Computer Interaction Laboratory of the Institute for Advanced Computer Studies and the Institute for Systems Research University of Maryland College Park MD
暂无评论