Other

Rohan Bopardikar

Sunil Keshav Bopardikar

Nikhil Sunil Bopardikar

Systems and methods for improving a machine learning model are described. In an embodiment, a computing system stores a plurality of training examples comprising training inputs and training outputs. The computing system generates a machine learning model and training the machine learning model using the plurality of training examples. The computing system receives a particular input for the machine learning system and, using the particular input and the machine learning system, computes a particular output. For each training example of the plurality of training examples, the computing system adjusts a weight of the training example on the machine learning system and computes a relative numerical impact on the particular output for the training example, the relative numerical impact reflecting an importance of each training example on the particular output relative to an importance of the other training examples of the plurality of training examples on the particular output. The server computer generates training example relevance data comprising identifiers of the plurality of training examples and the relative numerical impact values for the plurality of training examples. The server computer stores the training example relevance data in the computing system.

Identifying contributing training datasets for outputs of machine learning models

Systems and methods for reducing the size of neural networks are disclosed. In an embodiment, a server computer stores a plurality of training datasets, each of which comprise a plurality of training input matrices and a plurality of corresponding outputs. The server computer initiates training of a neural network using the plurality of training input matrices, a weight matrix, and the plurality of corresponding outputs. While the training of the neural network is being performed, the server computer identifies one or more weight values of the weight matrix for removal. The server computer removes the one or more weight values from the weight matrix to generate a reduced weight matrix. The server computer then stores the reduced weight matrix with the neural network.

Reducing the size of a neural network through reduction of the weight matrices

Systems and methods for reducing the size of deep neural networks are disclosed. In an embodiment, a server computer stores a plurality of training datasets, each of which comprise a plurality of training input matrices and a plurality of corresponding outputs. The server computer initiates training of a deep neural network using the plurality of training input matrices, a weight matrix, and the plurality of corresponding outputs. While the training of the deep neural network is being performed, the server computer identifies one or more weight values of the weight matrix for removal. The server computer removes the one or more weight values from the weight matrix to generate a reduced weight matrix. The server computer then stores the reduced weight matrix with the deep neural network.

This invention provides apparatuses, methods, and systems for classification of a web client's network bandwidth by a web server in real time over the Internet. The web server, based upon the round trip time (RTT) taken to establish the TCP connection with the web client, classifies the network bandwidth. The RTT for establishment of the TCP connection using a 3-way handshake is stored on the web server on most modern Operating Systems and can be fetched on demand by the web server for a given connection. A web application on the web server could then use this bandwidth classification to serve varied content to the web client, such as a light or heavy web page depending on the level of the bandwidth.

Growing community of inventors

Los Altos Hills, CA, United States of America

Rohan Bopardikar