Machine Learning in Java (book) (Other amazing things)
My second book "Machine Learning in Java" will provide you with the recipes and tools you need to quickly gain insight from complex data. You will learn how to apply machine learning methods to a variety of common tasks including classification, prediction, forecasting, market basket analysis, clustering, anomaly detection, recommendations, activity recognition, image recognition, and text analysis. By applying the most effective machine learning methods to real-world problems, you will gain hands-on experience that will transform the way you think about data. The book covers libraries such as Weka, Apache Mahout, Mallet, DeepLearning4java, and others.
Instant Weka How-to (book) (Other amazing things)
I authored the book Instant Weka How-to, which shows how to include Weka’s machinery in your Java application by implementing cutting-edge data-mining aspects such as regression and classification, forecasting, decision making, and recommendations.
The book starts by importing and preparing the data, and then moves on to more serious topics on classification, regression, clustering, and evaluation. The book also shows you how to implement online learning and how to create your own classifier. It includes several application examples such as house price prediction, stock value forecasting, and decision making for direct marketing.
Confidence: Ubiquitous Care System to Support Independent Living (Development)
The Confidence system aims at helping the elderly stay independent longer by detecting falls and unusual movement which may indicate a health problem. The system uses location sensors and wearable tags to determine the coordinates of the user’s body parts, and an accelerometer to detect fall impact and movement. Machine learning is combined with domain knowledge in the form of rules to recognize the user’s activity. The fall detection employs a similar combination of machine learning and domain knowledge. It was tested on five atypical falls and events that can be easily mistaken for a fall. While neither sensor type can correctly recognize all of these events on its own, the combination of both sensor types yields highly accurate fall detection. In addition, the detection of unusual movement can observe both the user’s micro-movement and macro-movement. This makes it possible to detect most types of threats to the user’s health and well-being manifesting in his/her movement.
Recommendation System for Mobile Turist Guide (Development)
The application first asks you where and when you want to go, and what kind of sights you are interested in. It then prepares the perfect itinerary for you, using artificial intelligence to learn your preferences from your ratings of the sights you have seen, as well as ratings of other users with similar tastes. Finally, it guides you on your trip, showing the sights on the map and providing written and spoken descriptions.
The application uses a hybrid recommender system based on expert knowledge, user ratings, and user profiles to filter relevant points of interest (POI). After that, it solves two coupled NP-complete problems, namely the knapsack problem and traveling salesman problem, to choose a subset of relevant POI and to find the optimal path between them.
Suspicious Behavior Detection (Development)
In many domains, no single observation event is sufficient to determine that the behavior is suspicious. Instead, suspiciousness must be inferred from a combination of multiple events, where events refer to the individual’s interactions with other individuals or environment. Hence, a detection system must employ a detector that combines evidence from multiple events unlike most previous work which focuses on the detection of a single, clearly suspicious event. We studied several detectors and proposed a novel F-UPR detector. The evaluation was performed in a simulated airport domain, where the goal is to catch a suspicious passenger that avoids security personnel while trying to achieve a secured point.
An Advanced Biometric Access Control System (Development)
Biometric and smart card access points are increasingly becoming equipped with additional input sensors, integrated intelligent video surveillance systems, and advanced intelligent methods that learn from experience. We have designed a flexible modular system based on integration of arbitrary access control sensors and an arbitrary number of stand-alone modules. The system was tested with four sensors (door sensor, identity card reader, fingerprint reader, camera) and four independent modules (expert-defined rules, micro learning, macro learning and visual learning). The designed prototype shows encouraging results and justifies integration of intelligent detection of unusual behavior in future access control systems for intelligent homes, industry facilities, offices, hospitals, military campuses, etc.