Tech stack

Ambient Intelligence Platform

 

The Ambient Intelligence Platform mines the data collected by the SDK for rich and understandable patterns of human behavior. The Platform is 100% cloud-based and provides deep, up-to-date analysis results at low latencies.

State-of-the-art signal processing and machine learning techniques detect and explain spatio-temporal patterns in your sensor data. Our algorithms provide highly accurate transport mode classification, venue mapping, map matching, home or work detection, semantic time modeling, and more.

The platform aggregates these events into relevant moment by means of unsupervised learning of temporal routines and expected behavior.
Real-time moment detections include a user’s commute, traveling and holiday moments, working remote moments, shopping and resto moments, etc.

Sensor fusion

State-of-the-art probabilistic sensor fusion techniques are used to merge together different data streams such as those coming from accelerometers, gyroscope, location subsystems, smart home motion sensors, etc.

Early sensor fusion allows our SDK to continuously estimate the phone’s orientation, separating longitudinal from latitudinal accelerations, and allows our platform to estimate a user’s indoor location based on a set of simple motion sensors.

The normalized data streams are fed into our ambient intelligence platform and used for event detection and device calibration.

Cognitive computing

Our deep recurrent neural network pipeline is able to predict expected events and moments before they happen, thereby guiding our anomaly detection framework.

A semantic embedding of these events, moments and anomalies is used to automatically segment the population and to determine the user’s long-term behavioral profile.

The combination of a user’s detected events, moments, anomalies and segments yields the user’s final context, representing a rich representation of the user’s current state, past behavior and predicted actions.

These data points are exposed through our profile and context stores and are easily accessible by means of a REST API. Moreover, our dashboards can be used to quickly visualize your population.

Mobile SDK

Available as Android and iOS libraries, optimized for easy integration into your mobile applications, it performs dynamic autonomous data collection from the smartphone’s sensors. Configurable to achieve a perfect balance between accuracy and battery usage for your specific use case.

Ambient detections

The SDK operates autonomously without user or developer interaction. Proprietary ambient sensing technology embedded in the SDK initiates and stops sensor data collection at the most opportune times. Data from locations, accelerometer, gyroscope and more is collected and serialized into low on-the-wire data sizes for minimal bandwidth usage.

API

The Sentiance API unlocks the power of Ambient Intelligence. The RESTful API allows for direct integration with any programming language or system with easy to process JSON resources. Extensive documentation and an interactive API explorer let you connect with the API in minutes. Offloads for bulk ingestion can be provided as well.

Designed for efficiency, built for accuracy.

Battery friendly through intelligent adaptive resource usage.
The SDK is battery friendly. This is achieved through intelligent design and advanced battery optimization whereby the SDK continuously adapts its mode of operation to consume a minimum of available resources on the device.