18341697. APPROXIMATE K NEAREST NEIGHBORS ON HARDWARE ACCELERATORS simplified abstract (GOOGLE LLC)
APPROXIMATE K NEAREST NEIGHBORS ON HARDWARE ACCELERATORS
Organization Name
Inventor(s)
Felix Ren-Chyan Chern of New York NY (US)
Blake Alan Hechtman of Mountain View CA (US)
Andrew Thomas Davis of Menlo Park CA (US)
Ruiqi Guo of Jersey City NJ (US)
Sanjiv Kumar of Jericho NY (US)
David Alexander Majnemer of Mountain View CA (US)
APPROXIMATE K NEAREST NEIGHBORS ON HARDWARE ACCELERATORS - A simplified explanation of the abstract
This abstract first appeared for US patent application 18341697 titled 'APPROXIMATE K NEAREST NEIGHBORS ON HARDWARE ACCELERATORS
Simplified Explanation
The patent application describes methods, systems, and apparatus for performing a kNN computation using a hardware accelerator. The abstract outlines the steps involved in the process, including obtaining query vectors and database vectors, and using a hardware accelerator to compute similarity values and identify the most similar database vectors for each query vector.
- The patent application focuses on performing a kNN computation using a hardware accelerator.
- The method involves obtaining a set of query vectors and a set of database vectors.
- The hardware accelerator is used to compute similarity values between each query vector and each database vector.
- For each query vector, the hardware accelerator identifies the index and similarity value of the most similar database vector within each bin.
Potential Applications
- This technology can be applied in various fields that require similarity-based computations, such as machine learning, data mining, and recommendation systems.
- It can be used to improve the efficiency and speed of kNN computations in large-scale datasets.
Problems Solved
- Traditional kNN computations can be computationally expensive and time-consuming, especially for large datasets.
- This technology solves the problem by utilizing a hardware accelerator to perform the computations, resulting in faster and more efficient processing.
Benefits
- The use of a hardware accelerator improves the speed and efficiency of kNN computations.
- It allows for faster processing of large-scale datasets, enabling real-time or near-real-time applications.
- The technology can be integrated into existing systems and workflows, enhancing their performance without significant modifications.
Original Abstract Submitted
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for performing a kNN computation using a hardware accelerator. One of the methods includes obtaining a set of one or more query vectors; obtaining a set of database vectors; and performing, on a hardware accelerator and for each query vector in the set, a search for the k most similar database vectors to the query vector, comprising: computing, by circuitry of the hardware accelerator and for each query vector, a respective similarity value between the query vector and each database vector; and for each query vector, identifying, by the hardware accelerator and for each bin, (i) an index of the most similar database vector within the bin and (ii) the respective similarity value for the most similar database vector within the bin.
