arXiv subjects

Medha Atre

Publications and source records attributed to Medha Atre.

7 recordsLinked to original sources

arXiv2022

Detecting Cloud-Based Phishing Attacks by Combining Deep Learning Models

Web-based phishing attacks nowadays exploit popular cloud web hosting services and apps such as Google Sites and Typeform for hosting their attacks. Since these attacks originate from reputable domains and IP addresses of the cloud services, traditional phishing detection methods such as IP reputation monitoring and blacklisting are not very effective. Here we investigate the effectiveness of deep learning models in detecting this class of cloud-based phishing attacks. Specifically, we evaluate deep learning models for three phishing detection methods--LSTM model for URL analysis, YOLOv2 model for logo analysis, and triplet network model for visual similarity analysis. We train the models using well-known datasets and test their performance on cloud-based phishing attacks in the wild. Our results qualitatively explain why the models succeed or fail. Furthermore, our results highlight how combining results from the individual models can improve the effectiveness of detecting cloud-based phishing attacks.

cs.CR↗

arXiv2021

Distributed Deep Learning Using Volunteer Computing-Like Paradigm

Use of Deep Learning (DL) in commercial applications such as image classification, sentiment analysis and speech recognition is increasing. When training DL models with large number of parameters and/or large datasets, cost and speed of training can become prohibitive. Distributed DL training solutions that split a training job into subtasks and execute them over multiple nodes can decrease training time. However, the cost of current solutions, built predominantly for cluster computing systems, can still be an issue. In contrast to cluster computing systems, Volunteer Computing (VC) systems can lower the cost of computing, but applications running on VC systems have to handle fault tolerance, variable network latency and heterogeneity of compute nodes, and the current solutions are not designed to do so. We design a distributed solution that can run DL training on a VC system by using a data parallel approach. We implement a novel asynchronous SGD scheme called VC-ASGD suited for VC systems. In contrast to traditional VC systems that lower cost by using untrustworthy volunteer devices, we lower cost by leveraging preemptible computing instances on commercial cloud platforms. By using preemptible instances that require applications to be fault tolerant, we lower cost by 70-90% and improve data security.

cs.DC↗

arXiv2020

Needles in the 'Sheet'stack: Augmented Analytics to get Insights from Spreadsheets

Business intelligence (BI) tools for database analytics have come a long way and nowadays also provide ready insights or visual query explorations, e.g. QuickInsights by Microsoft Power BI, SpotIQ by ThoughtSpot, Zenvisage, etc. In this demo, we focus on providing insights by examining periodic spreadsheets of different reports (aka views), without prior knowledge of the schema of the database or reports, or data information. Such a solution is targeted at users without the familiarity with the database schema or resources to conduct analytics in the contemporary way.

cs.DB↗

arXiv2019

Verity: Blockchains to Detect Insider Attacks in DBMS

Integrity and security of the data in database systems are typically maintained with access control policies and firewalls. However, insider attacks -- where someone with an intimate knowledge of the system and administrative privileges tampers with the data -- pose a unique challenge. Measures like append only logging prove to be insufficient because an attacker with administrative privileges can alter logs and login records to eliminate the trace of attack, thus making insider attacks hard to detect. In this paper, we propose Verity -- first of a kind system to the best of our knowledge. Verity serves as a dataless framework by which any blockchain network can be used to store fixed-length metadata about tuples from any SQL database, without complete migration of the database. Verity uses a formalism for parsing SQL queries and query results to check the respective tuples' integrity using blockchains to detect insider attacks. We have implemented our technique using Hyperledger Fabric, Composer REST API, and SQLite database. Using TPC-H data and SQL queries of varying complexity and types, our experiments demonstrate that any overhead of integrity checking remains constant per tuple in a query's results, and scales linearly.

cs.DB↗

arXiv2018

Algorithms and Analysis for the SPARQL Constructs

As Resource Description Framework (RDF) is becoming a popular data modelling standard, the challenges of efficient processing of Basic Graph Pattern (BGP) SPARQL queries (a.k.a. SQL inner-joins) have been a focus of the research community over the past several years. In our recently published work we brought community's attention to another equally important component of SPARQL, i.e., OPTIONAL pattern queries (a.k.a. SQL left-outer-joins). We proposed novel optimization techniques -- first of a kind -- and showed experimentally that our techniques perform better for the low-selectivity queries, and give at par performance for the highly selective queries, compared to the state-of-the-art methods. BGPs and OPTIONALs (BGP-OPT) make the basic building blocks of the SPARQL query language. Thus, in this paper, treating our BGP-OPT query optimization techniques as the primitives, we extend them to handle other broader components of SPARQL such as such as UNION, FILTER, and DISTINCT. We mainly focus on the procedural (algorithmic) aspects of these extensions. We also make several important observations about the structural aspects of complex SPARQL queries with any intermix of these clauses, and relax some of the constraints regarding the cyclic properties of the queries proposed earlier. We do so without affecting the correctness of the results, thus providing more flexibility in using the BGP-OPT optimization techniques.

cs.DB↗

arXiv2015

Left Bit Right: For SPARQL Join Queries with OPTIONAL Patterns (Left-outer-joins)

SPARQL basic graph pattern (BGP) (a.k.a. SQL inner-join) query optimization is a well researched area. However, optimization of OPTIONAL pattern queries (a.k.a. SQL left-outer-joins) poses additional challenges, due to the restrictions on the \textit{reordering} of left-outer-joins. The occurrence of such queries tends to be as high as 50% of the total queries (e.g., DBPedia query logs). In this paper, we present \textit{Left Bit Right} (LBR), a technique for \textit{well-designed} nested BGP and OPTIONAL pattern queries. Through LBR, we propose a novel method to represent such queries using a graph of \textit{supernodes}, which is used to aggressively prune the RDF triples, with the help of compressed indexes. We also propose novel optimization strategies -- first of a kind, to the best of our knowledge -- that combine together the characteristics of \textit{acyclicity} of queries, \textit{minimality}, and \textit{nullification}, \textit{best-match} operators. In this paper, we focus on OPTIONAL patterns without UNIONs or FILTERs, but we also show how UNIONs and FILTERs can be handled with our technique using a \textit{query rewrite}. Our evaluation on RDF graphs of up to and over one billion triples, on a commodity laptop with 8 GB memory, shows that LBR can process \textit{well-designed} low-selectivity complex queries up to 11 times faster compared to the state-of-the-art RDF column-stores as Virtuoso and MonetDB, and for highly selective queries, LBR is at par with them.

cs.DB↗

arXiv2012

BitPath -- Label Order Constrained Reachability Queries over Large Graphs

In this paper we focus on the following constrained reachability problem over edge-labeled graphs like RDF -- "given source node x, destination node y, and a sequence of edge labels (a, b, c, d), is there a path between the two nodes such that the edge labels on the path satisfy a regular expression "*a.*b.*c.*d.*". A "*" before "a" allows any other edge label to appear on the path before edge "a". "a.*" forces at least one edge with label "a". ".*" after "a" allows zero or more edge labels after "a" and before "b". Our query processing algorithm uses simple divide-and-conquer and greedy pruning procedures to limit the search space. However, our graph indexing technique -- based on "compressed bit-vectors" -- allows indexing large graphs which otherwise would have been infeasible. We have evaluated our approach on graphs with more than 22 million edges and 6 million nodes -- much larger compared to the datasets used in the contemporary work on path queries.

cs.DB↗