Federated doubly stochastic kernel learning on vertical partitioned data

A method for prediction using a machine learning model, comprising:

• obtaining, by an active computing device, parameters of the machine learning model;
• retrieving, by the active computing device, an instance from the local data of the active computing device;
• sampling, by the active computing device, a random direction of the instance;
• computing, by the active computing device, a dot product of the random direction and the instance, and calculating a random feature based on the dot product;
• computing, by the active computing device, a predicted value of the instance, instructing at least one passive computing device to compute a predicted value of the instance therein, and summarizing the predicted values from the active and the at least one passive computing devices to obtain a final predicted value of the instance, wherein the predicted value of the instance in the at least one passive computing device is obtained based on the local data of the at least one passive computing device;
• determining, by the active computing device, a model coefficient using the random feature and a difference between the final predicted value of the instance and a target value of the instance;
• updating, by the active computing device, the machine learning model using the model coefficient;
• and
• predicting, by the active computing device, a value for a new instance using the machine learning model.

A non-transitory computer readable medium storing computer executable code, wherein the computer executable code, when executed at a processor of an active computing device, is configured to:

• obtain parameters of a machine learning model;
• retrieve an instance from the local data of the active computing device;
• sample a random direction of the instance;
• compute a dot product of the random direction and the instance, and calculate a random feature based on the dot product;
• compute a predicted value of the instance in the active computing device, instruct at least one passive computing device to compute a predicted value of the instance therein, and summarize the predicted values from the active and the at least one passive computing devices to obtain a final predicted value of the instance, wherein the predicted value of the instance in the at least one passive computing device is obtained based on the local data of the at least one passive computing device;
• determine a model coefficient using the random feature and a difference between the final predicted value of the instance and a target value of the instance;
• update the machine learning model using the model coefficient;
• and
• predict a value for a new instance using the machine learning model.