Abstract

This paper introduces an efficient privacy-preserving protocol for distributed K-means clustering over an arbitrary partitioned data, shared among N parties. Clustering is one of the fundamental algorithms used in the field of data mining. Advances in data acquisition methodologies have resulted in collection and storage of vast quantities of user’s personal data. For mutual benefit, organizations tend to share their data for analytical purposes, thus raising privacy concerns for the users. Over the years, numerous attempts have been made to introduce privacy and security at the expense of massive additional communication costs. The approaches suggested in the literature make use of the cryptographic protocols such as Secure Multiparty Computation (SMC) and/or homomorphic encryption schemes like Paillier’s encryption. Methods using such schemes have proven communication overheads. And in practice are found to be slower by a factor of more than 106. In light of the practical limitations posed by privacy using the traditional approaches, we explore a paradigm shift to side-step the expensive protocols of SMC. In this work, we use the paradigm of secret sharing, which allows the data to be divided into multiple shares and processed separately at different servers. Using the paradigm of secret sharing, allows us to design a provably-secure, cloud computing based solution which has negligible communication overhead compared to SMC and is hence over a million times faster than similar SMC based protocols

Abstract

Computing convex hull for a given set of points is one of the most explored problems in the area of computational geometry (CG). If the set of points is distributed among a set of parties who jointly wish to compute the convex hull, each party can send his points to every other party, and can then locally compute the hull using any of the existing algorithms in CG. However such an approach does not work if the parties wish to compute the convex hull securely, i.e., no party wishes to reveal any of his input points to any other party apart from those that are part of the answer. The problem of secure computation of convex hull for two parties was first introduced by Du and Atallah (NSPW '01). The first solution to the problem was given by Wang et. al(ARES '08). However, the proposed solution was based on well known algorithms for computing convex hull in CG which are proven to be sub-optimal. We propose a new solution for secure computation of convex hull with a considerable improvement in computational complexity. We further show how to extend our two-party protocol for the case of any number of parties.

Abstract

In this paper, we re-visit the problem of unconditionally secure message transmission (USMT) from a sender S to a receiver R, who are part of a distributed synchronous network, modeled as an arbitrary directed graph. Some of the intermediate nodes between S and R can be under the control of an adversary having unbounded computing power. Desmedt and Wang [4] have given the characterization of USMT in directed networks. However, in their model, the underlying network is abstracted as directed node disjoint paths (also called as wires/channels) between S and R, where the intermediate nodes are oblivious, message passing nodes and perform no other computation. In this work, we first show that the characterization of USMT given by Desmedt et.al [4] does not hold good for arbitrary directed networks, where the intermediate nodes can perform some computation, beside acting as message forwarding nodes. We then give the characterization of USMT in arbitrary directed networks, considering the entire network as a whole. As far our knowledge is concerned, this is the first ever characterization of USMT in arbitrary directed networks.

Abstract

This paper proposes a method to cluster documents of variable length. The main idea is to apply (a) automatic identification of 1, 2, and 3 grams (To reduce the dependency on huge background vocabulary support or learning or complex probabilistic approach), (b) order them by some measure of relevance, which is developed with the help of Tf-Idf and Term-Weighting approach, and finally (c) use them (instead of bag of words based approach) to create vector space model and apply some known clustering methods i. e. Bisecting K-means, K-means, hierarchical method (single link) and Graph based method. Our experimental results with publicly available text dataset (Cogprints and NewsGroup20) show remarkable improvements in the performance of these clustering algorithms with this new approach.

Abstract

In the PSMT problem, a sender S and a receiver R are part of a distributed network and connected through n node disjoint paths, also called as wires among which at most t are controlled by an all powerful Byzantine adversary At. S has a message m, which S intends to send to R. The challenge is to design a protocol, such that at the end, R should correctly output m without any error (perfect reliability) and At should not get any information about m, what so ever, in information theoretic sense (perfect security). The problem of USMT is same as PSMT, except that R should output m with a small probability of error.Sayeed et al. [15] have given a PSMT protocol in an asynchronous network tolerating At, where S and R are connected by n = 2t + 1 wires. However, we show that their protocol does not provide perfect security. We then prove that in an asynchronous network, if all the n wires are directed from S to R, then any PSMT protocol tolerating At is possible iff n > 3t. Surprisingly, we further prove that even if all the n wires are bi-directional, then any PSMT protocol in asynchronous network tolerating At is possible iff n > 3t. This is quite interesting because for synchronous networks, by the results of Dolev et al. [6], if all the wires are unidirectional (directed from S to R), then PSMT tolerating At is possible iff n > 3t, where as if all the wires are bi-directional then PSMT tolerating At is possible iff n > 2t. This shows that synchrony of the network affects the connectivity requirement for PSMT protocols. However, we show that n > 2t wires are necessary and sufficient for the existence of any USMT protocol in asynchronous network tolerating At, irrespective of whether the n wires are unidirectional from S to R or the n wires are bi-directional.

Abstract

Semantic annotation or tagging of images can greatly improve the accuracy and efficiency of image search engines. However, humans rarely annotate images as they find the task of image annotation boring and laborious despites its benefits in terms of search and retrieval. In this paper, we introduce a novel approach of luring users into image annotation: by embedding image annotation into a CAPTCHA design. A CAPTCHA is a standard security mechanism used by popular commercial websites to prevent automated programs from abusing the online services. Millions of users solve CAPTCHAs daily in order to access web content and services. We aim to utilize human effort spent in solving the CAPTCHA into a productive work of image annotation. We introduce iCAPTCHA, a user friendly and productive CAPTCHA design. Our premise is based on the human ability to recognize images and label them in proper categories. Each time a user solves an iCAPTCHA, he/she is helping to label images in proper categories which will in turn improve image search and retrieval.

Abstract

The significant growth in computational power of modern Graphics Processing Units (GPUs) coupled with the advent of general purpose programming environments like NVIDIA's CUDA, has seen GPUs emerging as a very popular parallel computing platform. Till recently, there has not been a performance model for GPGPUs. The absence of such a model makes it difficult to definitively assess the suitability of the GPU for solving a particular problem and is a significant impediment to the mainstream adoption of GPUs as a massively parallel (super)computing platform. In this paper we present a performance prediction model for the CUDA GPGPU platform. This model encompasses the various facets of the GPU architecture like scheduling, memory hierarchy, and pipelining among others. We also perform experiments that demonstrate the effects of various memory access strategies. The proposed model can be used to analyze pseudo code for a CUDA kernel to obtain a performance estimate, in a way that is similar to performing asymptotic analysis. We illustrate the usage of our model and its accuracy with three case studies: matrix multiplication, list ranking, and histogram generation.

Abstract

Biometric authentication over public networks leads to a variety of privacy issues that needs to be addressed before it can become popular. The primary concerns are that the biometrics might reveal more information than the identity itself, as well as provide the ability to track users over an extended period of time. In this paper, we propose an authentication protocol that alleviates these concerns. The protocol takes care of user privacy, template protection and trust issues in biometric authentication systems. The protocol uses asymmetric encryption, and captures the advantages of biometric authentication. The protocol provides non-repudiable identity verification, while not revealing any additional information about the user to the server or vice versa. We show that the protocol is secure under various attacks. Experimental results indicate that the overall method is efficient to be used in practical scenarios.

Abstract

Widespread use of surveillance cameras in offices and other business establishments, pose a significant threat to the privacy of the employees and visitors. The challenge of introducing privacy and security in such a practical surveillance system has been stifled by the enormous computational and communication overhead required by the solutions. In this paper, we propose an efficient framework to carry out privacy preserving surveillance. We split each frame into a set of random images. Each image by itself does not convey any meaningful information about the original frame, while collectively, they retain all the information. Our solution is derived from a secret sharing scheme based on the Chinese Remainder Theorem, suitably adapted to image data. Our method enables distributed secure processing and storage, while retaining the ability to reconstruct the original data in case of a legal requirement. The system installed in an office like environment can effectively detect and track people, or solve similar surveillance tasks. Our proposed paradigm is highly efficient compared to Secure Multiparty Computation, making privacy preserving surveillance, practical.

Abstract

Secure multiparty protocols have found applications in numerous domains, where multiple nontrusting parties wish to evaluate a function of their private inputs. In this paper, we consider the case of multiple robots wishing to localize themselves, with maps as their private inputs. Though localization of robots has been a well studied problem, only recent studies have shown how to actively localize multiple robots through coordination. In all such studies, localization has typically been achieved through constructing a publicly known global map. Here, we show how a similar solution can be given in the case of nontrusting robots, which do not wish to disclose their local maps.