Page 32 - FDMAsia MarApr 2025

30 TECHNOLOGY MAR/APR 2025 FDM ASIA | www.fdmasia.com

features and a fixed number of decision trees are integrated
to achieve higher predictive performance and robustness.
The advantages of RF include:
(1) Handling large-scale, high-dimensional feature datasets

with powerful adaptability to disturbances in the dataset,
exhibiting high robustness in various ways.
(2) The relative importance of different features can be
determined by assessing the contribution of each input
variable in nonlinear sensitivity analysis and decision-making
processes.
(3) Adapting to application scenarios with incomplete data,
performing classification or regression in situations where
the given dataset is incomplete.

(4) Since the RF algorithm is a non-parametric model, it does
and objective aesthetic features. not make assumptions about the existence of any a priori
Aesthetic experience involves multiple sensory inputs, such relationships in the dataset and can deal with datasets
as vision, hearing and touch. Recent research in affective that contain non-linear relationships.
computing and user experience evaluation has highlighted
the value of multimodal measurement in assessing aesthetic This study used Nordic-style chairs as an example to
experiences. explore synchronous measurement methods of HCI applied

For instance, Kim and Lee developed a multimodal to the furniture domain. In addition, the impact of form and
framework that combines video, audio, text, and physiological colour on aesthetic preferences for chairs was investigated. The
data to improve emotion recognition and user satisfaction study aims to advance the integration of HCI techniques and
assessment, proving the value of multimodal integration for furniture design, particularly through synchronous measurement
robust user experience analysis. methods for user behaviour and emotional assessment.
Similarly, Wang et al. introduced an attention-enhanced To achieve this, we propose a multimodal measurement
model for emotion recognition in complex environments, method that includes questionnaires, eye tracking, physiological
using Transformers and attention mechanisms to capture measures, and EEG acquisition to investigate the factors
subtle emotional cues in multimedia settings, making it highly (i.e., colour and form of chairs, sex, and consumer specialty)

effective for understanding aesthetic and emotional responses. that significantly affect subjective preference evaluation and
Machine learning techniques can effectively process physiological feedback.
multimodal data, integrate different sensory inputs, and improve And we also established an aesthetic model using RF
the model’s ability to comprehensively understand and analyse algorithm based on the multimodal measurement data. This
the aesthetic experience. approach provides a clear and objective perspective on
Breiman proposed the RF algorithm, which is an emerging furniture preferences.
and highly flexible machine learning algorithm used to effectively It benefits the research and exploration of innovative furniture

address classification and regression problems. The core idea design and imparts a sense of visualisation, accuracy, and
of the RF algorithm is ensemble learning, where different scientific rigor, enhancing its overall significance.