Debajyoti Bera

Small depth quantum circuits

Small depth quantum circuits have proved to be unexpectedly powerful in comparison to their classical counterparts. We survey some of the recent work on this and present some open problems.

On the impact of seed scheduling in peer-to-peer networks

In a content distribution (file sharing) scenario, the initial phase is delicate due to the lack of global knowledge and the dynamics of the overlay. An unwise piece dissemination in this phase can cause delays in reaching steady state, thus increasing file download times. After showing that finding the scheduling strategy for optimal dissemination is computationally hard, even when the offline knowledge of the overlay is given, we devise a new class of scheduling algorithms at the seed (source peer with full content), based on a proportional fair approach, and we implement them on a real file sharing client. In addition to simulation results, we validated on our own file sharing client (BUTorrent) that our solution improves up to 25% the average downloading time of a standard file sharing protocol. Moreover, we give theoretical upper bounds on the improvements that our scheduling strategies may achieve.

Frequent-Itemset Mining Using Locality-Sensitive Hashing

The Apriori algorithm is a classical algorithm for the frequent itemset mining problem. A significant bottleneck in Apriori is the number of I/O operation involved, and the number of candidates it generates. We investigate the role of LSH techniques to overcome these problems, without adding much computational overhead. We propose randomized variations of Apriori that are based on asymmetric LSH defined over Hamming distance and Jaccard similarity.

Identification of Major Signaling Pathways in Prion Disease Progression Using Network Analysis.

Prion diseases are transmissible neurodegenerative diseases that arise due to conformational change of normal, cellular prion protein (PrPC) to protease-resistant isofrom (rPrPSc). Deposition of misfolded PrpSc proteins leads to an alteration of many signaling pathways that includes immunological and apoptotic pathways. As a result, this culminates in the dysfunction and death of neuronal cells. Earlier works on transcriptomic studies have revealed some affected pathways, but it is not clear which is (are) the prime network pathway(s) that change during the disease progression and how these pathways are involved in crosstalks with each other from the time of incubation to clinical death. We perform network analysis on large-scale transcriptomic data of differentially expressed genes obtained from whole brain in six different mouse strain-prion strain combination models to determine the pathways involved in prion diseases, and to understand the role of crosstalks in disease propagation. We employ a notion of differential network centrality measures on protein interaction networks to identify the potential biological pathways involved. We also propose a crosstalk ranking method based on dynamic protein interaction networks to identify the core network elements involved in crosstalk with different pathways. We identify 148 DEGs (differentially expressed genes) potentially related to the prion disease progression. Functional association of the identified genes implicates a strong involvement of immunological pathways. We extract a bow-tie structure that is potentially dysregulated in prion disease. We also propose an ODE model for the bow-tie network. Predictions related to diseased condition suggests the downregulation of the core signaling elements (PI3Ks and AKTs) of the bow-tie network. In this work, we show using transcriptomic data that the neuronal dysfunction in prion disease is strongly related to the immunological pathways. We conclude that these immunological pathways occupy influential positions in the PFNs (protein functional networks) that are related to prion disease. Importantly, this functional network involvement is prevalent in all the five different mouse strain-prion strain combinations that we studied. We also conclude that the dysregulation of the core elements of the bow-tie structure, which belongs to PI3K-Akt signaling pathway, leads to dysregulation of the downstream components corresponding to other biological pathways.

Efficient Parallel Ear Decomposition of Graphs with Application to Betweenness-Centrality

Parallel graph algorithms continue to attract a lotof research attention given their applications to several fields ofsciences and engineering. Efficient design and implementation ofgraph algorithms on modern manycore accelerators has to how-ever contend with a host of challenges including not being able toreach full memory system throughput and irregularity. Of late, focusing on real-world graphs, researchers are addressing thesechallenges by using decomposition and preprocessing techniquesguided by the structural properties of such graphs. In this direction, we present a new GPU algorithm forobtaining an ear decomposition of a graph. Our implementationof the proposed algorithm on an NVidia Tesla K40c improvesthe state-of-the-art by a factor of 2.3x on average on a collectionof real-world and synthetic graphs. The improved performanceof our algorithm is due to our proposed characterization thatidentifies edges of the graph as redundant for the purposes of anear decomposition. We then study an application of the ear decomposition of agraph in computing the betweenness-centrality values of nodesin the graph. We use an ear decomposition of the input graph tosystematically remove nodes of degree two. The actual computation of betweenness-centrality is done on the remaining nodesand the results are extended to nodes removed in the previousstep. We show that this approach improves the state-of-the-artfor computing betweenness-centrality on an NVidia K40c GPUby a factor of 1.9x on average over a collection of real-world graphs.

Efficient Enforcement of Privacy for Moving Object Trajectories

Information services based on identity and current location is already very popular among Internet and Mobile users, and a recent trend that is gaining acceptance is those based on annotated routes of travel, which we call as trajectories. We are motivated by the need of some users to reveal neither their identity nor location. This is not impossible since exact location can be substituted by an enclosing region, and the identity can be anonymised by relaying all queries through a proxy. However, when users are continuously making queries throughout a session, their queries can contain sufficient correlation which can identify them and/or their queries. Furthermore, a large region will fetch unnecessary search results degrading search quality. This problem of guaranteeing privacy, using smallest possible enclosing regions is NP-hard in general. We propose an efficient greedy algorithm which guarantees a user specified level of location and query privacy, namely k-anonymity and l-diversity, throughout a session and all the while trying to not significantly compromise service quality. Our algorithm, running on the proxy, makes use of trajectories to find similar users whose trajectories are also close by using appropriate notions of similarity and closeness for privacy enforcement. We give an indexing structure for efficiently storing and retrieving past trajectories, and present extensive experimental results comparing our approach with other similar approaches.

Preserving location privacy for continuous queries on known route

Protecting privacy in location based services has recently received considerable attention. Various approaches have been proposed, ranging from mix-zones to cloaking. Cloaking based approaches are ill-suited for continuous queries, where correlation between regular location updates may disclose location information. We consider the cloaking strategy with a modification to suit continuous queries: skip location updates at some key positions. The objective is to trade service availability at some locations in exchange of privacy at all times. Considering the case where the entire path of the user is known in advance, we show how to strategically decide these locations in a manner which is efficient, and does not skip too many locations (compared to the optimum). Experimental results show the validity and effectiveness of the proposed algorithm.

A lower bound method for quantum circuits

Quantum circuits, which are shallow, limited in the number of gates and additional workspace qubits, are popular for quantum computation because they form the simplest possible model similar to the classical model of a network of Boolean gates and capable of performing non-trivial computation. We give a new lower bound technique for such circuits and use it to give another proof that deterministic computation of the parity function cannot be performed by such circuits.

Detection and Diagnosis of Single Faults in Quantum Circuits

Detection and isolation of faults is a crucial step in the physical realization of quantum circuits. Even though quantum gates and circuits compute reversible functions, the standard techniques of automatic test pattern generation (ATPG) for classical reversible circuits are not directly applicable to quantum circuits. For faulty quantum circuits under the widely accepted single fault assumption, we show that their behavior can be fully characterized by the (single) faulty gate and the corresponding fault model. This allows us to efficiently determine test input states as well as measurement strategy for fault detection and diagnosis. Building on top of these, we design randomized algorithms which are able to detect every nontrivial single-gate fault with minimal probability of error. We also describe similar algorithms for fault diagnosis. We evaluate our algorithms by the number of output samples that needs to be collected and the probability of error. Both of these can be related to the eigenvalues of the operators corresponding to the circuit gates. We experimentally compare all our strategies with the state-of-the-art ATPG techniques for quantum circuits under the “single missing faulty gate” model and demonstrate that significant improvement is possible if we can exploit the quantum nature of circuits.

A hybrid framework for mining high-utility itemsets in a sparse transaction database

High-utility itemset mining aims to find the set of items with utility no less than a user-defined threshold in a transaction database. High-utility itemset mining is an emerging research area in the field of data mining and has important applications in inventory management, query recommendation, systems operation research, bio-medical analysis, etc. Currently, known algorithms for this problem can be classified as either 1-phase or 2-phase algorithms. The 2-phase algorithms typically consist of tree-based algorithms which generate candidate high-utility itemsets and verify them later. A tree data structure generates candidate high-utility itemsets quickly by storing some upper bound utility estimate at each node. The 1-phase algorithms typically consist of inverted-list based and transaction projection based algorithms which avoid the generation of candidate high-utility itemsets. The inverted list and transaction projection allows computation of exact utility estimates. We propose a novel hybrid framework that combines a tree-based and an inverted-list based algorithm to efficiently mine high-utility itemsets. Algorithms based on the framework can harness benefits of both types of algorithms. We report experiment results on real and synthetic datasets to demonstrate the effectiveness of our framework.