WO2016186826A1 - Rotating data for neural network computations - Google Patents

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for computing a layer output for a convolutional neural network layer, the method comprising: receiving a plurality of activation inputs; forming a plurality of vector inputs from the plurality of activation inputs, each vector input comprising values from a distinct region within the multi-dimensional matrix; sending the plurality of vector inputs to one or more cells along a first dimension of the systolic array; generating a plurality of rotated kernel structures from each of the plurality of kernel; sending each kernel structure and each rotated kernel structure to one or more cells along a second dimension of the systolic array; causing the systolic array to generate an accumulated output based on the plurality of value inputs and the plurality of kernels; and generating the layer output from the accumulated output.

Date: 2020/02/10 15:01

Related Entries

Read more [1801.00746] Bridging the Gap Between Neural Networks and Neuromorphic Hardware with A Neural Networ...
0 users, 1 mentions 2020/03/03 08:20
Read more ACM AI | Intro to Machine Learning: Advanced Track Workshops | CS
0 users, 1 mentions 2020/03/07 12:54
Read more Microsoft, Google and WIMI’s Deep Learning Accelerates the Development of AI - Press Release -...
0 users, 1 mentions 2020/05/18 20:21
Read more [2004.04509] A Survey on Coarse-Grained Reconfigurable Architectures from a Performance Perspectiveo...
0 users, 1 mentions 2020/09/21 15:51
Read more [2009.12495] Rubik: A Hierarchical Architecture for Efficient Graph Learningopen searchopen navigati...
0 users, 1 mentions 2020/09/29 02:22