Inferring High-Level Behavior from Low-Level Sensors
نویسندگان
چکیده
We present a method of learning a Bayesian model of a traveler moving through an urban environment. This technique is novel in that it simultaneously learns a unified model of the traveler’s current mode of transportation as well as his most likely route, in an unsupervised manner. The model is implemented using particle filters and learned using Expectation-Maximization. The training data is drawn from a GPS sensor stream that was collected by the authors over a period of three months. We demonstrate that by adding more external knowledge about bus routes and bus stops, accuracy is improved.
منابع مشابه
Inferring High-Level Behavior from Low-Level Sensors DRAFT — Not For Citation!
We present a method of learning a Bayesian model of a traveller moving through an urban environment. This technique is novel in that it simultaneously learns a unified model of the traveller’s current mode of transportation as well as his most likely route, in an unsupervised manner. The model is implemented using particle filters and learned using Expectation-Maximization. The training data is...
متن کاملInferring Complex Human Behavior Using a Non-obtrusive Mobile Sensing Platform
Thanks to the decreasing cost, increasing mobility, and wider use of sensors, a great number of possible applications have recently emerged, including applications that may impact the high-level goals in people’s lives. Applications can be found in many areas ranging from medical devices to consumer devices. Information about activity and context can be inferred from sensors and be used to prov...
متن کاملCapturing Individual and Group Behavior with Wearable Sensors
We show how to obtain high level descriptions of human behavior in terms of physical activity, speech activity, face-to-face interaction (f2f), physical proximity, and social network attributes from sensor data. We present experimental results that show that it is possible to identify individual personality traits as well as subjective and objective group performance metrics from low level data...
متن کاملSynthetic Character with Bayesian Network and Behavior Network for Intelligent Smartphone
As more people get using mobile phones, smartphone which is a new generation of mobile phone with computing capability earns world-wide reputation as new personal business assistant and entertainment equipment. This paper presents a synthetic character which acts as a user assistant and an entertainer in smartphone. It collects low-level information from various information sources available in...
متن کاملInteractively Inferring Intent against Driver Models Constrained by Metric Temporal Logic
This work addresses the problem of inferring the high-level behavior of a decision-making agent when the specific model governing its actions is unknown; only constraints on candidate models, in the form of metric temporal logic formulae, are known. We develop a worst-case tree search solution framework and apply it to the case of validating driver intent in multi-vehicle pursuit scenarios.
متن کامل