skip to main content

Search for: All records

Creators/Authors contains: "Soundarajan, Sucheta"

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. We present EASEE (Edge Advertisements into Snapshots using Evolving Expectations) for partitioning streaming communication data into static graph snapshots. Given streaming communication events (A talks to B), EASEE identifies when events suffice for a static graph (a snapshot ). EASEE uses combinatorial statistical models to adaptively find when a snapshot is stable, while watching for significant data shifts – indicating a new snapshot should begin. If snapshots are not found carefully, they poorly represent the underlying data – and downstream graph analytics fail: We show a community detection example. We demonstrate EASEE's strengths against several real-world datasets, and its accuracy against known-answer synthetic datasets. Synthetic datasets' results show that (1) EASEE finds known-answer data shifts very quickly; and (2) ignoring these shifts drastically affects analytics on resulting snapshots. We show that previous work misses these shifts. Further, we evaluate EASEE against seven real-world datasets (330 K to 2.5B events), and find snapshot-over-time behaviors missed by previous works. Finally, we show that the resulting snapshots' measured properties (e.g., graph density) are altered by how snapshots are identified from the communication event stream. In particular, EASEE's snapshots do not generally “densify” over time, contradicting previous influential results that used simpler partitioning methods.
    Free, publicly-accessible full text available December 1, 2023
  2. In many network applications, it may be desirable to conceal certain target nodes from detection by a data collector, who is using a crawling algorithm to explore a network. For example, in a computer network, the network administrator may wish to protect those computers (target nodes) with sensitive information from discovery by a hacker who has exploited vulnerable machines and entered the network. These networks are often protected by hiding the machines (nodes) from external access, and allow only fixed entry points into the system (protection against external attacks). However, in this protection scheme, once one of the entry points is breached, the safety of all internal machines is jeopardized (i.e., the external attack turns into an internal attack). In this paper, we view this problem from the perspective of the data protector. We propose the Node Protection Problem: given a network with known entry points, which edges should be removed/added so as to protect as many target nodes from the data collector as possible? A trivial way to solve this problem would be to simply disconnect either the entry points or the target nodes – but that would make the network non-functional. Accordingly, we impose certain constraints: for eachmore »node, only (1 − r) fraction of its edges can be removed, and the resulting network must not be disconnected. We propose two novel scoring mechanisms - the Frequent Path Score and the Shortest Path Score. Using these scores, we propose NetProtect, an algorithm that selects edges to be removed or added so as to best impede the progress of the data collector. We show experimentally that NetProtect outperforms baseline node protection algorithms across several real-world networks. In some datasets, With 1% of the edges removed by NetProtect, we found that the data collector requires up to 6 (4) times the budget compared to the next best baseline in order to discover 5 (50) nodes.« less
  3. Throughout many scientific and engineering fields, including control theory, quantum mechanics, advanced dynamics, and network theory, a great many important applications rely on the spectral decomposition of matrices. Traditional methods such as the power iteration method, Jacobi eigenvalue method, and QR decomposition are commonly used to compute the eigenvalues and eigenvectors of a square and symmetric matrix. However, these methods suffer from certain drawbacks: in particular, the power iteration method can only find the leading eigen-pair (i.e., the largest eigenvalue and its corresponding eigenvector), while the Jacobi and QR decomposition methods face significant performance limitations when facing with large scale matrices. Typically, even producing approximate eigenpairs of a general square matrix requires at least O(N^3) time complexity, where N is the number of rows of the matrix. In this work, we exploit the newly developed memristor technology to propose a low-complexity, scalable memristor-based method for deriving a set of dominant eigenvalues and eigenvectors for real symmetric non-negative matrices. The time complexity for our proposed algorithm is O(N^2 /Δ) (where Δ governs the accuracy). We present experimental studies to simulate the memristor-supporting algorithm, with results demonstrating that the average error for our method is within 4%, while its performance is upmore »to 1.78X better than traditional methods.« less
  4. The high computation and memory storage of large deep neural networks (DNNs) models pose intensive challenges to the conventional Von-Neumann architecture, incurring substantial data movements in the memory hierarchy. The memristor crossbar array has emerged as a promising solution to mitigate the challenges and enable low-power acceleration of DNNs. Memristor-based weight pruning and weight quantization have been separately investigated and proven effectiveness in reducing area and power consumption compared to the original DNN model. However, there has been no systematic investigation of memristor-based neuromorphic computing (NC) systems considering both weight pruning and weight quantization. In this paper, we propose an unified and systematic memristor-based framework considering both structured weight pruning and weight quantization by incorporating alternating direction method of multipliers (ADMM) into DNNs training. We consider hardware constraints such as crossbar blocks pruning, conductance range, and mismatch between weight value and real devices, to achieve high accuracy and low power and small area footprint. Our framework is mainly integrated by three steps, i.e., memristor- based ADMM regularized optimization, masked mapping and retraining. Experimental results show that our proposed frame- work achieves 29.81× (20.88×) weight compression ratio, with 98.38% (96.96%) and 98.29% (97.47%) power and area reduction on VGG-16 (ResNet-18) networkmore »where only have 0.5% (0.76%) accuracy loss, compared to the original DNN models. We share our models at anonymous link http://bit.ly/2Jp5LHJ .« less