arXiv subjects

Kristof Van Laerhoven

Publications and source records attributed to Kristof Van Laerhoven.

40 records · Page 3Linked to original sources

arXiv2021

On-site Online Feature Selection for Classification of Switchgear Actuations

As connected sensors continue to evolve, interest in low-voltage monitoring solutions is increasing. This also applies in the area of switchgear monitoring, where the detection of switch actions, their differentiation and aging are of fundamental interest. In particular, the universal applicability for various types of construction plays a major role. Methods in which design-specific features are learned in an offline training are therefore less suitable for assessing the condition of switchgears. A new computational efficient method for intelligent online feature selection is presented, which can be used to train a model for the addressed use cases on-site. Process- and design-specific features can be learned locally (e.g. on a sensor system) without the need of prior offline training. The proposed method is evaluated on four datasets of switchgear measurements, which were recorded using microelectromechanical system (MEMS) based sensors (acoustic and vibration). Furthermore, we show that the features selected by our method can be used to track changes in switching processes due to aging effects.

eess.SY↗

arXiv2019

A Multi-Media Exchange Format for Time-Series Dataset Curation

Exchanging data as character-separated values (CSV) is slow, cumbersome and error-prone. Especially for time-series data, which is common in Activity Recognition, synchronizing several independently recorded sensors is challenging. Adding second level evidence, like video recordings from multiple angles and time-coded annotations, further complicates the matter of curating such data. A possible alternative is to make use of standardized multi-media formats. Sensor data can be encoded in audio format, and time-coded information, like annotations, as subtitles. Video data can be added easily. All this media can be merged into a single container file, which makes the issue of synchronization explicit. The incurred performance overhead by this encoding is shown to be negligible and compression can be applied to optimize storage and transmission overhead.

cs.DB↗

arXiv2018

Wearable affect and stress recognition: A review

Affect recognition aims to detect a person's affective state based on observables, with the goal to e.g. provide reasoning for decision making or support mental wellbeing. Recently, besides approaches based on audio, visual or text information, solutions relying on wearable sensors as observables (recording mainly physiological and inertial parameters) have received increasing attention. Wearable systems offer an ideal platform for long-term affect recognition applications due to their rich functionality and form factor. However, existing literature lacks a comprehensive overview of state-of-the-art research in wearable-based affect recognition. Therefore, the aim of this paper is to provide a broad overview and in-depth understanding of the theoretical background, methods, and best practices of wearable affect and stress recognition. We summarise psychological models, and detail affect-related physiological changes and their measurement with wearables. We outline lab protocols eliciting affective states, and provide guidelines for ground truth generation in field studies. We also describe the standard data processing chain, and review common approaches to preprocessing, feature extraction, and classification. By providing a comprehensive summary of the state-of-the-art and guidelines to various aspects, we would like to enable other researchers in the field of affect recognition to conduct and evaluate user studies and develop wearable systems.

cs.HC↗

arXiv2015

An Automated System for Discovering Neighborhood Patterns in Ego Networks

Generally, social network analysis has often focused on the topology of the network without considering the characteristics of individuals involved in them. Less attention is given to study the behavior of individuals, considering they are the basic entity of a graph. Given a mobile social network graph, what are good features to extract key information from the nodes? How many distinct neighborhood patterns exist for ego nodes? What clues does such information provide to study nodes over a long period of time? In this report, we develop an automated system in order to discover the occurrences of prototypical ego-centric patterns from data. We aim to provide a data-driven instrument to be used in behavioral sciences for graph interpretations. We analyze social networks derived from real-world data collected with smart-phones. We select 13 well-known network measures, especially those concerned with ego graphs. We form eight feature subsets and then assess their performance using unsupervised clustering techniques to discover distinguishing ego-centric patterns. From clustering analysis, we discover that eight distinct neighborhood patterns have emerged. This categorization allows concise analysis of users' data as they change over time. The results provide a fine-grained analysis for the contribution of different feature sets to detect unique clustering patterns. Last, as a case study, two datasets are studied over long periods to demonstrate the utility of this method. The study shows the effectiveness of the proposed approach in discovering important trends from data.

cs.SI↗