Click For Photo: https://regmedia.co.uk/2019/05/15/search_computer.jpg
Microsoft has open sourced its machine learning algorithm that powers its search engine Bing.
“Keyword search algorithms just fail when people ask a question or take a picture and ask the search engine, ‘What is this?’” said Rangan Majumder, group program manager on Microsoft’s Bing search and AI team.
Queries - Text - Microsoft - Engineers - Pieces
To avoid matching search queries by text only, Microsoft engineers have encoded over 150 billion pieces of data in different formats like words, web pages or images as vectors. Representing data in this way is common in deep learning, as it helps neural networks learn the underlying patterns in a particular dataset.
The vectors used by Bing engineers aren’t fed into a neural network as input, however. Instead, another machine learning algorithm known as Space Partition Tree And Graph (SPTAG) searches through these vectors. It clusters similar vectors together and finds the closest neighboring vectors that are related to what is being searched to give the most relevant information.
Example - Someone - Image - Eiffel - Tower
For example, if someone searches an image of the Eiffel Tower, Bing will convert that picture into a vector using a PyTorch model and then the SPTAG algorithm finds...
Wake Up To Breaking News!
Warum denn nicht?