skip to main content

Search for: All records

Creators/Authors contains: "Zhou, Z"

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. Free, publicly-accessible full text available May 1, 2023
  2. Free, publicly-accessible full text available April 21, 2023
  3. Free, publicly-accessible full text available January 1, 2023
  4. Abstract—Hyperdimensional Computing (HDC) is a neurallyinspired computation model working based on the observation that the human brain operates on high-dimensional representations of data, called hypervector. Although HDC is significantly powerful in reasoning and association of the abstract information, it is weak on features extraction from complex data such as image/video. As a result, most existing HDC solutions rely on expensive pre-processing algorithms for feature extraction. In this paper, we propose StocHD, a novel end-to-end hyperdimensional system that supports accurate, efficient, and robust learning over raw data. Unlike prior work that used HDC for learning tasks, StocHD expands HDC functionality tomore »the computing area by mathematically defining stochastic arithmetic over HDC hypervectors. StocHD enables an entire learning application (including feature extractor) to process using HDC data representation, enabling uniform, efficient, robust, and highly parallel computation. We also propose a novel fully digital and scalable Processing In-Memory (PIM) architecture that exploits the HDC memorycentric nature to support extensively parallel computation. Our evaluation over a wide range of classification tasks shows that StocHD provides, on average, 3.3x and 6.4x (52.3x and 143.Sx) faster and higher energy efficiency as compared to state-of-the-art HDC algorithm running on PIM (NVIDIA GPU), while providing 16x higher computational robustness.« less
    Free, publicly-accessible full text available December 1, 2022
  5. Abstract

    Eukaryotic flagella (synonymous with cilia) rely on a microtubule-based axoneme, together with accessory filaments to carryout motility and signaling functions. While axoneme structures are well characterized, 3D ultrastructure of accessory filaments and their axoneme interface are mostly unknown, presenting a critical gap in understanding structural foundations of eukaryotic flagella. In the flagellum of the protozoan parasiteTrypanosoma brucei(T. brucei), the axoneme is accompanied by a paraflagellar rod (PFR) that supports non-planar motility and signaling necessary for disease transmission and pathogenesis. Here, we employed cryogenic electron tomography (cryoET) with sub-tomographic averaging, to obtain structures of the PFR, PFR-axoneme connectors (PACs), andmore »the axonemal central pair complex (CPC). The structures resolve how the 8 nm repeat of the axonemal tubulin dimer interfaces with the 54 nm repeat of the PFR, which consist of proximal, intermediate, and distal zones. In the distal zone, stacked “density scissors” connect with one another to form a “scissors stack network (SSN)” plane oriented 45° to the axoneme axis; and ~370 parallel SSN planes are connected by helix-rich wires into a paracrystalline array with ~90% empty space. Connections from these wires to the intermediate zone, then to overlapping layers of the proximal zone and to the PACs, and ultimately to the CPC, point to a contiguous pathway for signal transmission. Together, our findings provide insights into flagellum-driven, non-planar helical motility ofT. bruceiand have broad implications ranging from cell motility and tensegrity in biology, to engineering principles in bionics.

    « less
  6. Solar module recycling is unprofitable today. In this paper potential revenues from waste Si modules are analyzed. The biggest revenue potential comes from the Si cells, extracted intact or broken. The second revenue source is the bulky materials in the modules including Al frame, Cu wiring and glass. The total revenue is estimated between US$11–30/module depending on the percentage of cells extracted intact. This revenue is 4–10 times better than today’s recycling process that recovers only the bulky materials. Experimentally a special furnace has been demonstrated to successfully separate thin commercial Si cells of 160 microns from glass unbroken. Frommore »damaged cells a new chemistry has been developed to recover solar-grade Si and Ag. It requires fewer steps than today’s recycling process, with Ag recovery of 97% and Si recovery of 90%. A prototype recycling line is needed to assess the cost of the new process.« less