Classification of examples by multiple agents with private features

We consider classification tasks where relevant features are distributed among a set of agents and cannot be centralized, for example due to privacy restrictions. We are motivated by a key classification task that arises in a calendar management domain where software assistants classify new meetings as likely to be difficult to schedule. Accurate prediction of the output class is difficult for an isolated single agent because the target concept may involve features to which the agent does not have access, for example each attendee's willingness to attend the meeting. To increase prediction accuracy, novel learning algorithms are required in which agents collaborate to classify new examples while maintaining the privacy of features. We introduce a novel distributed asynchronous decision-tree inspired algorithm for such tasks named DDT. DDT differs from previous approaches in that it applies to vertically partitioned data with categorical multi-valued features, it requires no explicit hypothesis generation, and there is no a priori restriction on number of agents. We present empirical results in our meeting scheduling domain and show that DDT outperforms a single agent learner and performs as well as a centralized learner with hypothetical access to all the features.