In this paper, we face the Affective Movement Recognition Challenge 2021 which is based on 3 naturalistic datasets on body movement, which is a fundamental component of everyday living both in the execution of the actions that make up physical functioning as well as in rich expression of affect, cognition, and intent. The datasets were built on deep understanding of the requirements of automatic detection technology for chronic pain physical rehabilitation, maths problem solving, and interactive dance contexts respectively. In particular, we will rely on a single, simple yet effective, approach able to be competitive with state-of-the-art results in the literature on all of the 3 datasets. Our approach is based on a two step procedure: first we will carefully handcraft features able to fully and synthetically represent the raw data and then we will apply Random Forest and XGBoost, carefully tuned with rigorous statistical procedures, on top of it to deliver the predictions. As requested by the challenge, we will report results in terms of three different metrics: accuracy, F1-score, and Matthew Correlation Coefficient.

Keep it Simple: Handcrafting Feature and Tuning Random Forests and XGBoost to face the Affective Movement Recognition Challenge 2021

D'Amato V.;Oneto L.;Camurri A.;Anguita D.
2021

Abstract

In this paper, we face the Affective Movement Recognition Challenge 2021 which is based on 3 naturalistic datasets on body movement, which is a fundamental component of everyday living both in the execution of the actions that make up physical functioning as well as in rich expression of affect, cognition, and intent. The datasets were built on deep understanding of the requirements of automatic detection technology for chronic pain physical rehabilitation, maths problem solving, and interactive dance contexts respectively. In particular, we will rely on a single, simple yet effective, approach able to be competitive with state-of-the-art results in the literature on all of the 3 datasets. Our approach is based on a two step procedure: first we will carefully handcraft features able to fully and synthetically represent the raw data and then we will apply Random Forest and XGBoost, carefully tuned with rigorous statistical procedures, on top of it to deliver the predictions. As requested by the challenge, we will report results in terms of three different metrics: accuracy, F1-score, and Matthew Correlation Coefficient.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11567/1086606
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact