Category Archives: ECE

Saeid Nooshabadi is Co-author of Article in Journal of Parallel and Distributed Computing

Saeid Nooshabadi

Saeid Nooshabadi (SAS/ECE), professor of electrical and computer engineering, is co-author of the article, “High-dimensional image descriptor matching using highly parallel KD-tree construction and approximate nearest neighbor search,” to be published in  the October 2019 issue of the Journal of Parallel and Distributed Computing, which is published by Elsevier. The article is co-authored by Michigan Tech Computer Science department doctoral candidate Linjia Hu.

Abstract: To overcome the high computational cost associated with the high-dimensional digital image descriptor matching, this paper presents a set of integrated parallel algorithms for the construction of K-dimensional tree (KD-tree) and P approximate nearest neighbor search (P-ANNS) on the modern massively parallel architectures (MPA). To improve the runtime performance of the P-ANNS, we propose an efficient sliding window for a parallel buffered P-ANNS on KD-tree to mitigate the high cost of global memory accesses. When applied to high dimensional real-world image descriptor datasets, the proposed KD-tree construction and the buffered P-ANNS algorithms are of comparable matching quality to the traditional sequential counterparts on CPU, while outperforming their serial CPU counterparts by speedup factors of up to 17 and 163, respectively. The algorithms are also studied for the performance impact factors to obtain the optimal runtime configurations for various datasets. Moreover, we verify the features of the parallel algorithms on typical 3D image matching scenarios. With the classical local image descriptor signature of histograms of orientations (SHOT) datasets, the parallel KD-tree construction and image descriptor matching can achieve up to 11 and 138-fold speedups, respectively.

Citation: Hu, L., & Nooshabadi, S. (2019). High-dimensional image descriptor matching using highly parallel KD-tree construction and approximate nearest neighbor search. Journal of Parallel and Distributed Computing, 132, 127-140. http://dx.doi.org/10.1016/j.jpdc.2019.06.003

MTU Digital Commons link: https://digitalcommons.mtu.edu/michigantech-p/145/

Elsevier link: https://www.sciencedirect.com/science/article/pii/S0743731519304319?via%3Dihub


Zhou Feng is PI on $500K NSF Project

Zhuo Feng (ECE/ICC) is Principal Investigator on a project that has received a $500,000 research and development grant from the National Science Foundation. This potential three-year project is titled, “SHF: Small: Spectral Reduction of Large Graphs and Circuit Networks.”

This research project will investigate a truly-scalable yet unified spectral graph reduction approach that allows reducing large-scale, real-world directed and undirected graphs with guaranteed preservation of the original graph spectra. Unlike prior methods that are only suitable for handling specific types of graphs (e.g. undirected or strongly-connected graphs), this project uses a more universal approach and thus will allow for spectral reduction of a much wider range of real-world graphs that may involve billions of elements:

  • spectrally-reduced social (data) networks allow for more efficiently modeling, mining and analysis of large social (data) networks;
  • spectrally-reduced neural networks allow for more scalable model training and processing in emerging machine learning tasks;
  • spectrally-reduced web-graphs allow for much faster computations of personalized PageRank vectors;
  • spectrally-reduced integrated circuit networks will lead to more efficient partitioning, modeling, simulation, optimization and verification of large chip designs, etc.

From Tech Today, June 21, 2019


Tim Havens Is Co-author of Article Published in IEEE Transactions on Fuzzy Systems

Timothy HavensTim Havens (CS/ICC) coauthored the article, “Enabling Explainable Fusion in Deep Learning with Fuzzy Integral Neural Networks,” which was accepted for publication in the journal IEEE Transactions on Fuzzy Systems.

Citation: M.A. Islam, D.T. Anderson, A. Pinar, T.C. Havens, G. Scott, and J.M. Keller. Enabling explainable fusion in deep learning with fuzzy integral neural networks. Accepted, IEEE Trans. Fuzzy Systems.

Abstract: Information fusion is an essential part of numerous engineering systems and biological functions, e.g., human cognition. Fusion occurs at many levels, ranging from the low-level combination of signals to the high-level aggregation of heterogeneous decision-making processes. While the last decade has witnessed an explosion of research in deep learning, fusion in neural networks has not observed the same revolution. Specifically, most neural fusion approaches are ad hoc, are not understood, are distributed versus localized, and/or explainability is low (if present at all). Herein, we prove that the fuzzy Choquet integral (ChI), a powerful nonlinear aggregation function, can be represented as a multi-layer network, referred to hereafter as ChIMP. We also put forth an improved ChIMP (iChIMP) that leads to a stochastic gradient descent-based optimization in light of the exponential number of ChI inequality constraints. An additional benefit of ChIMP/iChIMP is that it enables eXplainable AI (XAI). Synthetic validation experiments are provided and iChIMP is applied to the fusion of a set of heterogeneous architecture deep models in remote sensing. We show an improvement in model accuracy and our previously established XAI indices shed light on the quality of our data, model, and its decisions.