14.4 A 21.5M-query-vectors/s 3.37nJ/vector reconfigurable k-nearest-neighbor accelerator with adaptive precision in 14nm tri-gate CMOS.
Himanshu KaulMark A. AndersSanu K. MathewGregory K. ChenSudhir SatpathySteven HsuAmit AgarwalRam KrishnamurthyPublished in: ISSCC (2016)
Keyphrases
- k nearest neighbor
- knn
- vector space
- cmos technology
- spatial queries
- nearest neighbor
- query point
- nm technology
- low cost
- range queries
- metal oxide semiconductor
- feature vectors
- text categorization
- knn classifier
- support vector machine
- neural network
- database
- knn algorithm
- similarity search
- floating gate
- low power
- k nearest
- query processing
- vowel recognition
- distance computation
- power consumption
- high speed
- feature selection
- field programmable gate array
- graph construction
- text classification
- efficient processing
- distance function
- input vector
- distance measure
- multi dimensional
- decision trees
- low voltage
- training set
- knn classification
- data sets