Abstract
In this work, we introduce the MuSer, a propositional framework that explores temporal information available in multi-relational databases. At the core of this system is an encoding technique that translates the temporal information into a propositional sequence of events. By using this technique, we are able to explore the temporal information using a propositional sequence miner. With this framework, we mine each class partition individually and we do not use classical aggregation strategies, like window aggregation. Moreover, in this system we combine feature selection and propositionalization techniques to cast a multi-relational classification problem into a propositional one. We empirically evaluate the MuSer framework using two real databases. The results show that mining each partition individually is a time-and memory-efficient strategy that generates a high number of highly discriminative patterns.

Abstract

Abstract

Abstract

Abstract
Adverse Drug Events (ADEs) are a major health problem, and developing accurate prediction methods may have a significant impact in public health. Ideally, we would like to have predictive methods, that could pinpoint possible ADRs during the drug development process. Unfortunately, most relevant information on possible ADRs is only available after the drug is commercially available. As a first step, we propose using prior information on existing interactions through recommendation systems algorithms. We have evaluated our proposal using data from the ADReCS database with promising results.

Abstract
Markov Logic is an expressive and widely used knowledge representation formalism that combines logic and probabilities, providing a powerful framework for inference and learning tasks. Most Markov Logic implementations perform inference by transforming the logic representation into a set of weighted propositional formulae that encode a Markov network, the ground Markov network. Probabilistic inference is then performed over the grounded network. Constructing, simplifying, and evaluating the network are the main steps of the inference phase. As the size of a Markov network can grow rather quickly, Markov Logic Network (MLN) inference can become very expensive, motivating a rich vein of research on the optimization of MLN performance. We claim that parallelism can have a large role on this task. Namely, we demonstrate that widely available Graphics Processing Units (GPUs) can be used to improve the performance of a state-of-the-art MLN system, Tuffy, with minimal changes. Indeed, comparing the performance of our GPU-based system, TuGPU, to that of the Alchemy, Tuffy and RockIt systems on three widely used applications shows that TuGPU is up to 15x times faster than the other systems.