Sandip Shankarrao Patil

EDUCATION

Ph.D. in Computer Engineering

RESEARCH, TEACHING, or OTHER INTERESTS

Computer Engineering

Scopus Publications

Path and Information Content-Based Structural Word Sense Disambiguation
Sandip S. Patil, R. P. Bhavsar, and B. V. Pawar
Springer International Publishing



Detection of paraphrases for Devanagari languages using support vector machine
Darshana S. Bhole and Sandip S. Patil
IEEE
Paraphrase is a process of computing the semantic similarity between sentences, which are not lexicographically similar. It relates to the writing a sentence in another form. Though a number of metrics for English language have been proposed in literature, to quantify textual similarity; but none for Devanagari language. Existing system for Indian language paraphrase detection uses lexical similarity are supervised and requires large scale tagged corpus. The proposed method employs SVM learning metrics, based on lexicography similarity with producing output as +1 for paraphrased, −1 for not paraphrased, takes a sentence as input and produces another sentence without changing its semantic. In particular, the system addresses the problem for detection of monolingual text to text similarity for fusion language like Hindi and Marathi, which has complex morphology.


Wet and dry fingerprint enhancement by using multi resolution technique
Mayur S. Patil and Sandip S. Patil
IEEE
Biometrics is one of traditional and effective method for verification and validation. While using biometrics for security purpose, user does not need to remember password. Fingerprints are commonly used for authentication as they are different for every person, even twins have diverse fingerprints. Sensors for fingerprint take less space and flexible to use. With all above advantages there are some pitfalls, sometimes system may fail in authentication if condition of finger during verification is wrinkle and during enrollment is dry. This problem is generally faced by the people working in maritime environment. In such situations genuine person gets unauthenticated or vice versa and the system performance gets degraded. It is the problem in Public Distribution System as well as in Smart Vault Biometric System of banks. Existing system uses Gabor Filter, while in propose system the Rotational Filter is used.


Devloping knowledge driven ontology for decision making
Ashutosh V. Girase, Girish Kumar Patnaik, and Sandip S. Patil
IEEE
Decision-making activity is carried out in an every organization to solve the problems or to take the decisions. To make an effective decision, decision maker needs relevant and meaningful information. But to retrieve meaningful information from such a huge database, decision maker needs background knowledge about the domain. Practically for a decision maker having background knowledge about each and every domain is not possible. Due to this meaningful information remains hidden in the database itself. Decisions made out using such irrelevant and meaningless information leads to irreparable damage and harm to the reputation of organization. Hence to retrieve the meaningful and relevant information, background knowledge about the domain is necessary. To solve this problem, ontology is used as a source of domain knowledge. By using ontology, meaningful information is retrieved from the database to help in taking decision. In proposed system, ontology is used to represent the background knowledge about the domain. Use of ontology improves the relevancy and meaningfulness of the retrieved information. Such meaningful and relevant information is used to improve the effectiveness of decision making. Experimental analysis of the results shows that, results obtained by using proposed approach are more meaningful and relevant as compared to existing approach.


Improvement of data object's membership by using Fuzzy K-Means clustering approach
Chetana T. Baviskar and Sandip S. Patil
IEEE
Clustering is classified into two categories namely-Hard clustering and Soft clustering. The hard clustering restricts that the data object in the given data belongs to exactly one cluster. The problem with hard K-Means (KM) clustering is that the different initial partitions can result in different final clusters. Soft clustering which also known as fuzzy clustering forms clusters such that data object can belong to more than one cluster based on their membership values. But sometimes the resulting membership values do not always correspond well to the degrees of belonging of the data. So to overcome the problems in hard K-means (KM) clustering, the Fuzzy K-Means (FKM) clustering approach is proposed. The Proposed Fuzzy K-Means clustering assigns membership to an object inversely related to the relative distance of the object to cluster prototype. Fuzzy clustering uses membership values to assign data objects to one or more clusters. The membership values indicate the strength of the association between that data object and a particular cluster. The proposed work also compares the execution time and required memory of Proposed Fuzzy K-Means (FKM) to that of hard K-means (KM) clustering. The result shows membership of data object is improved, also the execution time and memory required for Proposed Fuzzy K-Means (FKM) clustering is less than that of hard K-Means (KM) clustering.


Selective review of fingerprint enhancement, classification and matching techniques
S. Patil Sandip and P. H. Zope
IEEE
Dry and wet fingerprint degrades the performance of fingerprint verification system, There are various issues in dry and wet fingerprint verification, like fingerprint skin quality, missing features and minimizing the matching efforts. Therefore, it is important for a dry and wet fingerprint recognition system to enhance, classify and match the fingerprint. In this work, we reviewed selective approaches for fingerprint enhancement, classification and matching.


Design and implementation of anomalies detection system using ip gray space analysis
Yogendra Kumar Jain and Sandip S. Patil
IEEE
In Network Security, there is a major issue to secure the public or private network from abnormal users. It is because each network is made up of users, services and computers with a specific behavior that is also called as heterogeneous system. To detect abnormal users, anomaly detection system (ADS) is used. In this paper, we present an Anomaly Detection System with the uses of IP gray space analysis and dominant scanning port identification heuristics used to detect various anomalous users with their potential behaviors. This methodology detects five types of anomalies with their potential behaviors and generates respective messages to GUI.


Tracking and identification of suspicious and abnormal behaviors using supervised machine learning technique
K. P. Adhiya, S. R. Kolhe, and Sandip S. Patil
ACM
The explosion in popularity of open systems interconnected via the pubic network and private network has made computer security an issue of increasing concern. In networks there is a major issue to secure the public or private network from suspicious behaviors. This is because each network is made of users, services and computers with a specific behavior that is then reflected in the generated network traffic to detect abnormal user suspicious detection system is used. This paper presents a supervised learning technique which is used to detect various suspicions and abnormal behaviors in the public or private networks. Our technique also detect various behaviors of these hosts using supervised learning methodology.

RECENT SCHOLAR PUBLICATIONS