Skip to Main Content Area
  • CPS-VO
    • Contact Support
  • Browse
    • Calendar
    • Announcements
    • Repositories
    • Groups
  • Search
    • Search for Content
    • Search for a Group
    • Search for People
    • Search for a Project
    • Tagcloud
      
 
Not a member?
Click here to register!
Forgot username or password?
 
Home
National Science Foundation

Cyber-Physical Systems Virtual Organization

Read-only archive of site from September 29, 2023.

CPS-VO

model parallelism

biblio

Visible to the public An Application of Storage-Optimal MatDot Codes for Coded Matrix Multiplication: Fast k-Nearest Neighbors Estimation

Submitted by grigby1 on Fri, 05/22/2020 - 2:39pm
  • parallel processing
  • faster query execution
  • high memory footprint
  • in-memory indexes
  • K-NN search
  • MatDot codes
  • matrix multiplication
  • model parallelism
  • model-parallel architecture
  • MRPT
  • multiple random projection trees
  • nearest neighbor estimation
  • fast k-nearest neighbors
  • query response time
  • query time
  • single compute node
  • single processing node
  • storage constraints
  • storage-optimal MatDot Codes
  • storage-optimal recovery threshold
  • straggler and failure tolerance
  • straggler-prone environment
  • Systematics
  • worker nodes
  • nearest neighbor search
  • Indexes
  • Data models
  • Computational modeling
  • Big Data
  • trees (mathematics)
  • Metrics
  • Measurement
  • nearest neighbour methods
  • data storage
  • query processing
  • mathematics computing
  • pubcrawl
  • high-dimensional data
  • Approximation algorithms
  • approximate nearest neighbor algorithms
  • candidate points
  • coded computing
  • coded matrix multiplication
  • computation efforts
  • computation strategy
  • distributed computing cluster
  • distributed processing
  • Euclidean distance calculations

Terms of Use  |  ©2023. CPS-VO