Section: New Results

Introduction

In this section, we organize the bulk of our contributions this year along two of our research axes, namely Pattern Mining and Decision Support. Some other contributions lie within the domain of machine learning.

Pattern Mining

In the domain of pattern mining we can categorize our contributions along the following lines:

• Efficient Pattern Mining (Sections 7.2-7.4). In [9], we propose a method to accelerate itemset sampling on FPGAs, whereas [18] proposes SSDPS, an efficient algorithm to mine discriminant patterns in two-class datasets, common in genetic data. Finally [11] presents a succinct data structure that represents concisely a cube of skypatterns.

• Semantics of Pattern Mining (Sections 7.5-7.6).[14] discusses the ambiguity of the semantics of pattern mining with absent events (negated statements). Likewise [8] shows formal properties of admissible generalizations in pattern mining and machine learning.

Decision Support

In regards to the axis of decision support, our contributions can be organized in two categories: forecasting & prediction, and modelisation.

• Forecasting & Prediction (Sections 7.7-7.9). In [10], we propose solutions to automate the task of capacity planning in the context of a large data network as the one available at Orange. [17] applies machine learning techniques for estrus detection in diary farms. [21] proposes a machine learning architecture in multi-sensor environments for earthquake early warning.

• Modelling (Section 7.10). In [5] we present a modeling approach for the nutritional requirements of lactating sows.

• Data Exploration (Section 7.11).[6] proposes a formal framework for the exploration of care trajectories in medical databases.

Others

• Machine Learning (Section 7.12-7.14).[7], [16] proposes novel methods to optimize the F-measure in ML, and to improve the task of domain adaptation by source selection. [19] proposes the use of GANs to make time series classification more interpretable.