Ishwar Sethi | Oakland University (original) (raw)

Papers by Ishwar Sethi

Research paper thumbnail of Mining association rules between low-level image features and high-level concepts

… of the SPIE Data Mining and …, 2001

Research paper thumbnail of Image retrieval using hierarchical self-organizing feature maps

Pattern Recognition Letters, 1999

This paper presents a scheme for image retrieval that lets a user retrieve images either by explo... more This paper presents a scheme for image retrieval that lets a user retrieve images either by exploring summary views of the image collection at diÄerent levels or by similarity retrieval using query images. The proposed scheme is based on image clustering through a hierarchy of self-organizing feature maps. While the suggested scheme can work with any kind of low-level feature

Research paper thumbnail of Hausdorff Metric Based Vector Quantization of Binary Images

Research paper thumbnail of Media content management

... For example, visualizing the areas of highest activity during tennis matches is a result of o... more ... For example, visualizing the areas of highest activity during tennis matches is a result of object (player) tracking throughout the game. ... Other techniques have been applied in the compressed domain (Arman, Hsu & Chiu, 1993; Gunsel, Ferman, & Tekalp, 1996; Shen, Li, & Sethi ...

Research paper thumbnail of Data mining: an introducation

Research paper thumbnail of Cardiac tamponade in systemic lupus erythematosus

Research paper thumbnail of Developing Audio Processing Agents for Multi-Agent MPEG-7 Enabled Environment

This paper presents a methodology for developing audio processing agents for a multi-agent enviro... more This paper presents a methodology for developing audio processing agents for a multi-agent environment that is known as the Community of Multimedia Agents. The Community's philosophy, objectives and architecture are described. The methodology is illustrated using audio feature extraction agents as example. The algorithms used for extracting audio features are classical and work in general audio domain. The agents have the standard MPEG-7 interface for better interoperation and wide usage. Two low level tools -the MPEG-7 audio descriptor wrapper classes and the MPEG audio decoder – are also presented. An example of agent aggregation for an annotation system prototyping is provided.

Research paper thumbnail of The PERSEUS Project: Creating Personalized Multimedia News Portal

This paper describes the Perseus project, which is devoted to developing techniques and tools for... more This paper describes the Perseus project, which is devoted to developing techniques and tools for creating personalized multimedia news portals. The purpose of a personalized multimedia news portal is to provide relevant information, selected from newswire sites on the Internet and augmented by video clips automatically extracted from TV broadcasts, based on the user's preferences. To create such an intelligent information system several techniques related to textual information retrieval, audio and video segmentation, and topic detection should be developed to work in accord. The approaches to event mining and tracking on the Internet, commercial detection and recognition in video and audio streams, and selection of relevant news video fragments, based on closed captioning and audio transcripts, are described.

Research paper thumbnail of PERSEUS: Personalized multimedia news portal

This paper describes the Perseus project, which is devoted to developing techniques and tools for... more This paper describes the Perseus project, which is devoted to developing techniques and tools for creating personalized multimedia news portals. The purpose of a personalized multimedia news portal is to provide releva nt information, selected from newswire sites on the Internet and augmented by video clips automatically extracted from TV broadcasts, based on the user's preferences. To create such an intelligent information system several techniques related to textual information retrieval, audio and video segmentation, and topic detection should be developed to work in accord. The approaches to event mining and tracking on the Internet, commercial detection and recognition in video and audio streams, and selection of relevant news video fragments, based on closed captioning and audio transcripts, are described.

Research paper thumbnail of Renal function status in leprosy

Indian Journal of Dermatology Venereology and Leprology

75 leprosy patients comprising of 47 males and 28 females ranging in age from 10 years to 75 year... more 75 leprosy patients comprising of 47 males and 28 females ranging in age from 10 years to 75 years (mean 33.76 years) were studied. Duration of disease ranged from one month to 12 years. Thirty (40%) cases were in reaction. Renal involvement occurred in 53.66% of cases. 80% of type 2 reaction leprosy patients showed renal involvement as compared to 53% of type 1 reaction. 80% of leprosy cases with more than 3 years duration of disease showed renal involvement as compared to 46.6% of cases with less than 3 years duration of disease. 9% of cases showed reduction in one or both kidney size with loss of corticomedullary differentiation on USG.

Research paper thumbnail of A Novel Distance Measure for Interval Data

Research paper thumbnail of A VLSI optimal constructive algorithm for classification problems

If neural networks are to be used on a large scale, they have to be implemented in hardware. Howe... more If neural networks are to be used on a large scale, they have to be implemented in hardware. However, the cost of the hardware implementation is critically sensitive to factors like the precision used for the weights, the total number of bits of information and the maximum fan-in used in the network. This paper presents a version of the Constraint

Research paper thumbnail of Iterative Split Adjustment for Building Multilabel Decision Trees

Research paper thumbnail of Neural implementation of tree classifiers

IEEE Transactions on Systems, Man, and Cybernetics, 1995

Tree classifiers represent a popular non-parametric classification methodology that has been succ... more Tree classifiers represent a popular non-parametric classification methodology that has been successfully used in many pattern recognition and learning tasks. However, “is feature-value⩾thrsh” type of tests used in tree classifiers are often found sensitive to noise and minor variations in the data. This has led to the use of soft thresholding in decision trees. Following the decision tree to feedforward

Research paper thumbnail of PART I: OVERVIEW OF DATA MINING

Research paper thumbnail of Motion filtering for target extraction and tracking in infrared images

A critical problem in automatic target recognition is the extraction and tracking of moving targe... more A critical problem in automatic target recognition is the extraction and tracking of moving targets in high clutter, low resolution, low contrast thermal imagery. This problem is further exacerbated by the large distances that typically exist between the sensor and the targets. In this paper, a new paradigm called motion filtering is presented for exploiting motion to extract and track targets. The main feature of the present approach is the emphasis on the motion information. Experimental results from two sequences of tactical situations are presented to exemplify the capabilities of the proposed approach.

Research paper thumbnail of Extraction of diagnostic rules using neural networks

A neural learning methodology is presented that is capable of providing rules from learned weight... more A neural learning methodology is presented that is capable of providing rules from learned weights. An application of this methodology to systematic lupus erythematosus is demonstrated. It is shown that the proposed approach can disregard irrelevant features in the data and can generate different criteria combinations indicating the presence of systemic lupus erythematosus in a patient

Research paper thumbnail of Multivalued Logic Mapping of Neurons in Feedforward Networks

A common view of feedforward neural networks is that of a black box since the knowledge embedded ... more A common view of feedforward neural networks is that of a black box since the knowledge embedded in the connection weights of a feedforward neural network is generally considered incomprehensible. Many researchers have addressed this deficiency of neural networks by suggesting schemes to obtain a Boolean logic representation for the output of a neuron based on its connection weights. However, these schemes mostly assume binary inputs to the neural network. Since it is not uncommon to find multivalued discrete inputs to neurons, we present in this paper a weight mapping scheme that is capable of generating a multivalued logic representation for the output of a neuron. Such a logic representation is also useful for continuous inputs through multilevel quantization. Two examples are presented to illustrate the use of multivalued logic representation in understanding the knowledge incorporated in the connection strengths of neurons in feedforward networks.

Research paper thumbnail of SIR: simultaneous induction of rules using neural networks

One major drawback of the decision-tree-based inductive knowledge acquisition methodology is its ... more One major drawback of the decision-tree-based inductive knowledge acquisition methodology is its inability to form high-level features from raw attributes. While neural learning has no such problem, its difficulty is in the opaqueness of the acquired knowledge. The authors address both these issues and present a neural learning methodology that yields production rules formed on the basis of high-level features that are also learned during the learning phase. Furthermore, the competitive component of the learning in the proposed methodology automatically determines the number of rules for a given learning situation. Two examples are presented to illustrate the methodology

Research paper thumbnail of Symbolic approximation of feedforward networks

Multiple layer f eedforward n eural networks are often v iewed as black box es as the knowledge s... more Multiple layer f eedforward n eural networks are often v iewed as black box es as the knowledge stored in the c onnection weights of these networks is generally considered incomprehensible. This paper suggests a solution to this deficiency o f neural networks by proposing a backtracking tree search procedure for converting the weights of a neuron into a symbolic representation and demonstrating its use for understanding and symbolic approximation of feedforward neural networks. Several examples are presented to illustrate the proposed symbolic mapping of neurons.

Research paper thumbnail of Mining association rules between low-level image features and high-level concepts

… of the SPIE Data Mining and …, 2001

Research paper thumbnail of Image retrieval using hierarchical self-organizing feature maps

Pattern Recognition Letters, 1999

This paper presents a scheme for image retrieval that lets a user retrieve images either by explo... more This paper presents a scheme for image retrieval that lets a user retrieve images either by exploring summary views of the image collection at diÄerent levels or by similarity retrieval using query images. The proposed scheme is based on image clustering through a hierarchy of self-organizing feature maps. While the suggested scheme can work with any kind of low-level feature

Research paper thumbnail of Hausdorff Metric Based Vector Quantization of Binary Images

Research paper thumbnail of Media content management

... For example, visualizing the areas of highest activity during tennis matches is a result of o... more ... For example, visualizing the areas of highest activity during tennis matches is a result of object (player) tracking throughout the game. ... Other techniques have been applied in the compressed domain (Arman, Hsu & Chiu, 1993; Gunsel, Ferman, & Tekalp, 1996; Shen, Li, & Sethi ...

Research paper thumbnail of Data mining: an introducation

Research paper thumbnail of Cardiac tamponade in systemic lupus erythematosus

Research paper thumbnail of Developing Audio Processing Agents for Multi-Agent MPEG-7 Enabled Environment

This paper presents a methodology for developing audio processing agents for a multi-agent enviro... more This paper presents a methodology for developing audio processing agents for a multi-agent environment that is known as the Community of Multimedia Agents. The Community's philosophy, objectives and architecture are described. The methodology is illustrated using audio feature extraction agents as example. The algorithms used for extracting audio features are classical and work in general audio domain. The agents have the standard MPEG-7 interface for better interoperation and wide usage. Two low level tools -the MPEG-7 audio descriptor wrapper classes and the MPEG audio decoder – are also presented. An example of agent aggregation for an annotation system prototyping is provided.

Research paper thumbnail of The PERSEUS Project: Creating Personalized Multimedia News Portal

This paper describes the Perseus project, which is devoted to developing techniques and tools for... more This paper describes the Perseus project, which is devoted to developing techniques and tools for creating personalized multimedia news portals. The purpose of a personalized multimedia news portal is to provide relevant information, selected from newswire sites on the Internet and augmented by video clips automatically extracted from TV broadcasts, based on the user's preferences. To create such an intelligent information system several techniques related to textual information retrieval, audio and video segmentation, and topic detection should be developed to work in accord. The approaches to event mining and tracking on the Internet, commercial detection and recognition in video and audio streams, and selection of relevant news video fragments, based on closed captioning and audio transcripts, are described.

Research paper thumbnail of PERSEUS: Personalized multimedia news portal

This paper describes the Perseus project, which is devoted to developing techniques and tools for... more This paper describes the Perseus project, which is devoted to developing techniques and tools for creating personalized multimedia news portals. The purpose of a personalized multimedia news portal is to provide releva nt information, selected from newswire sites on the Internet and augmented by video clips automatically extracted from TV broadcasts, based on the user's preferences. To create such an intelligent information system several techniques related to textual information retrieval, audio and video segmentation, and topic detection should be developed to work in accord. The approaches to event mining and tracking on the Internet, commercial detection and recognition in video and audio streams, and selection of relevant news video fragments, based on closed captioning and audio transcripts, are described.

Research paper thumbnail of Renal function status in leprosy

Indian Journal of Dermatology Venereology and Leprology

75 leprosy patients comprising of 47 males and 28 females ranging in age from 10 years to 75 year... more 75 leprosy patients comprising of 47 males and 28 females ranging in age from 10 years to 75 years (mean 33.76 years) were studied. Duration of disease ranged from one month to 12 years. Thirty (40%) cases were in reaction. Renal involvement occurred in 53.66% of cases. 80% of type 2 reaction leprosy patients showed renal involvement as compared to 53% of type 1 reaction. 80% of leprosy cases with more than 3 years duration of disease showed renal involvement as compared to 46.6% of cases with less than 3 years duration of disease. 9% of cases showed reduction in one or both kidney size with loss of corticomedullary differentiation on USG.

Research paper thumbnail of A Novel Distance Measure for Interval Data

Research paper thumbnail of A VLSI optimal constructive algorithm for classification problems

If neural networks are to be used on a large scale, they have to be implemented in hardware. Howe... more If neural networks are to be used on a large scale, they have to be implemented in hardware. However, the cost of the hardware implementation is critically sensitive to factors like the precision used for the weights, the total number of bits of information and the maximum fan-in used in the network. This paper presents a version of the Constraint

Research paper thumbnail of Iterative Split Adjustment for Building Multilabel Decision Trees

Research paper thumbnail of Neural implementation of tree classifiers

IEEE Transactions on Systems, Man, and Cybernetics, 1995

Tree classifiers represent a popular non-parametric classification methodology that has been succ... more Tree classifiers represent a popular non-parametric classification methodology that has been successfully used in many pattern recognition and learning tasks. However, “is feature-value⩾thrsh” type of tests used in tree classifiers are often found sensitive to noise and minor variations in the data. This has led to the use of soft thresholding in decision trees. Following the decision tree to feedforward

Research paper thumbnail of PART I: OVERVIEW OF DATA MINING

Research paper thumbnail of Motion filtering for target extraction and tracking in infrared images

A critical problem in automatic target recognition is the extraction and tracking of moving targe... more A critical problem in automatic target recognition is the extraction and tracking of moving targets in high clutter, low resolution, low contrast thermal imagery. This problem is further exacerbated by the large distances that typically exist between the sensor and the targets. In this paper, a new paradigm called motion filtering is presented for exploiting motion to extract and track targets. The main feature of the present approach is the emphasis on the motion information. Experimental results from two sequences of tactical situations are presented to exemplify the capabilities of the proposed approach.

Research paper thumbnail of Extraction of diagnostic rules using neural networks

A neural learning methodology is presented that is capable of providing rules from learned weight... more A neural learning methodology is presented that is capable of providing rules from learned weights. An application of this methodology to systematic lupus erythematosus is demonstrated. It is shown that the proposed approach can disregard irrelevant features in the data and can generate different criteria combinations indicating the presence of systemic lupus erythematosus in a patient

Research paper thumbnail of Multivalued Logic Mapping of Neurons in Feedforward Networks

A common view of feedforward neural networks is that of a black box since the knowledge embedded ... more A common view of feedforward neural networks is that of a black box since the knowledge embedded in the connection weights of a feedforward neural network is generally considered incomprehensible. Many researchers have addressed this deficiency of neural networks by suggesting schemes to obtain a Boolean logic representation for the output of a neuron based on its connection weights. However, these schemes mostly assume binary inputs to the neural network. Since it is not uncommon to find multivalued discrete inputs to neurons, we present in this paper a weight mapping scheme that is capable of generating a multivalued logic representation for the output of a neuron. Such a logic representation is also useful for continuous inputs through multilevel quantization. Two examples are presented to illustrate the use of multivalued logic representation in understanding the knowledge incorporated in the connection strengths of neurons in feedforward networks.

Research paper thumbnail of SIR: simultaneous induction of rules using neural networks

One major drawback of the decision-tree-based inductive knowledge acquisition methodology is its ... more One major drawback of the decision-tree-based inductive knowledge acquisition methodology is its inability to form high-level features from raw attributes. While neural learning has no such problem, its difficulty is in the opaqueness of the acquired knowledge. The authors address both these issues and present a neural learning methodology that yields production rules formed on the basis of high-level features that are also learned during the learning phase. Furthermore, the competitive component of the learning in the proposed methodology automatically determines the number of rules for a given learning situation. Two examples are presented to illustrate the methodology

Research paper thumbnail of Symbolic approximation of feedforward networks

Multiple layer f eedforward n eural networks are often v iewed as black box es as the knowledge s... more Multiple layer f eedforward n eural networks are often v iewed as black box es as the knowledge stored in the c onnection weights of these networks is generally considered incomprehensible. This paper suggests a solution to this deficiency o f neural networks by proposing a backtracking tree search procedure for converting the weights of a neuron into a symbolic representation and demonstrating its use for understanding and symbolic approximation of feedforward neural networks. Several examples are presented to illustrate the proposed symbolic mapping of neurons.