International Business Machines Corporation

John Donald Vasquez

Thiago Bianchi

Joao Souto Maior

John Maxwell Cohn

Ana Paula Brambila

An approach is provided in which the approach calculates at least one weighting factor based on a word frequency analysis of an unlabeled document against a set of word frequencies corresponding to a set of labeled documents. The approach computes an a posteriori classification probability of the unlabeled document based on the at least one weighting factor, and creates an inferred classifier based on the a posteriori classification probability. The approach classifies the unlabeled classifier using the inferred classifier.

Infer text classifiers for large text collections

Detecting motion-sensor data anomalies by registering an event associated with received motion-detector sensor data at a current timestamp, calculating, a timestamp delta between a previous timestamp and the current timestamp, and counting transitions between a set of timestamp deltas using a finite state machine. The anomaly detection also includes using a Markov chain transition matrix, generated from the finite state machine, to determine that, a probability of transitioning from a current state to a next state associated with the event, is less than a predetermined value, and registering an anomaly associated with the sensor and the event.

Detecting motion-sensor data anomalies

Monitoring indoor environmental conditions is provided. Sensor data and its corresponding time stamps from is collect from a number of environmental sensors within an enclosed environment. A set of all possible states is defined for a specified time period, each state representing a range of sensor data values. A probability of the system changing from any one state to another is modeled according to a Markov chain. When a new sensor data value is received from a sensor it is compared to a last sensor data value of a previous state, and a probability of transition from the previous state to the current state is determined. If the probability of transition from the previous state to the current state is less than a predetermined threshold, an anomaly is detected, and a service request is generated.

Growing community of inventors

Munich, Germany

John Donald Vasquez