QUANTIZATION OF ONLINE PRODUCTS USING APPROXIMATE NEAREST NEIGHBOUR SEARCH

Approximate nearest neighbor searching algorithms has achieved superior success in addition tasks. The existing well-liked methods for ANN search, like hashing and division. These methods are designed for static databases only. They cannot handle well we tend to address the matter by developing a web product division (online PQ) model and incrementally updating the division codebook that accommodates to the incoming streaming knowledge. Moreover, to additional alleviate the problem of large scale computation for the web PQ update; we tend to style to budget constraints for the model to update partial PQ codebook Instead of all. We tend to derive a loss sure that guarantees the performance of our on-line PQ model. Moreover, we tend to develop a web PQ model over a window with each knowledge insertion and deletion supported, to replicate the period behavior of the Data. The experiments demonstrate that our on-line PQ model is each time-efficient and effective for ANN search in dynamic giant scale databases compared with baseline strategies and also the plan of partial PQ codebook update additional reduces the update price