US Patent Application 17827222. SPARSIFYING VECTORS FOR NEURAL NETWORK MODELS BASED ON OVERLAPPING WINDOWS simplified abstract
Contents
SPARSIFYING VECTORS FOR NEURAL NETWORK MODELS BASED ON OVERLAPPING WINDOWS
Organization Name
Microsoft Technology Licensing, LLC
Inventor(s)
Girish Vishnu Varatkar of Sammamish WA (US)
Ankit More of San Mateo CA (US)
Bita Darvish Rouhani of Bellevue WA (US)
Mattheus C. Heddes of Redmond WA (US)
Gaurav Agrawal of San Jose CA (US)
SPARSIFYING VECTORS FOR NEURAL NETWORK MODELS BASED ON OVERLAPPING WINDOWS - A simplified explanation of the abstract
- This abstract for appeared for US patent application number 17827222 Titled 'SPARSIFYING VECTORS FOR NEURAL NETWORK MODELS BASED ON OVERLAPPING WINDOWS'
Simplified Explanation
This abstract describes systems and methods for reducing the size of vectors used in neural network models. It suggests using overlapping windows to select sets of elements from a vector. The highest absolute value element is then chosen from each set. The window is then moved along the vector by a specified number of elements, and the process is repeated to select another set of elements with at least one common element. The highest absolute value element from this set is also chosen. This approach helps to sparsify the vector by selecting only the most significant elements.
Original Abstract Submitted
Embodiments of the present disclosure include systems and methods for sparsifying vectors for neural network models based on overlapping windows. A window is used to select a first set of elements in a vector of elements. A first element is selected from the first set of elements having the highest absolute value. The window is slid along the vector by a defined number of elements. The window is used to select a second set of elements in the vector, wherein the first set of elements and the second set of elements share at least one common element. A second element is selected from the second set of elements having the highest absolute value.