Indonesian Journal of Electrical Engineering and Computer Science Vol. 38, No. 3, June 2025, pp. 1755~1764 ISSN: 2502-4752, DOI: 10.11591/ijeecs.v38.i3.pp1755-1764  1755 Journal homepage: http://ijeecs.iaescore.com Advancing SSVEP-based brain-computer interfaces: a novel approach using cross-subject multi-modal fusion technique Kalenahally R. Swetha, Ravikumar G. Krishnegowda, Shashikala S. Venkataramu Department of Computer Science and Engineering, BGSIT, Adichunchanagiri University, Mandya, India Article Info ABSTRACT Article history: Received Apr 30, 2024 Revised Nov 15, 2024 Accepted Nov 30, 2024 Brain-computer interfaces (BCIs) represent an innovative paradigm for device control and communication, relying solely on the analysis of brain activity. Steady-state visually evoked potentials (SSVEPs), characterized by neurophysiological responses synchronized with periodic visual stimuli, have gained prominence in BCI research due to their high information transfer rates (ITRs) and minimal user training requirements. However, the translation of SSVEP-based BCIs into practical applications faces challenges stemming from variations in user responses and stimuli. To address these issues, this study introduces a groundbreaking methodology known as the cross-subject multi-modal fusion technique (CMFT). CMFT revolutionizes template design by creating invariant templates resilient to user and stimulus differences, thereby enhancing SSVEP detection across diverse subjects and stimuli. The implications of this research extend to various fields, including assistive technologies, human-computer interaction, and cognitive neuroscience. CMFT presents a promising solution to make SSVEP-based BCIs more practical and widely applicable. The methodology involves intricate steps, including spatial filters, data pre-processing, and template generation, ensuring precise SSVEP detection. Through CMFT, this study contributes to advancing the effectiveness and versatility of SSVEP-based BCIs, fostering improved accessibility and interaction in a range of domains. Keywords: Brain-computer interface Cross-subject multi-modal fusion technique Electroencephalogram Information transfer rates Steady-state visually evoked potentials This is an open access article under the CC BY-SA license. Corresponding Author: Kalenahally R. Swetha Department of Computer Science and Engineering, BGSIT, Adichunchanagiri University Mandya, India Email: swethakr_12@rediffmail.com 1. INTRODUCTION Brain-computer interfaces (BCIs) offer a novel method of controlling equipment simply through the study of an individual's brain activity, which is accomplished by merging hardware and software components for successful communication [1]. Electroencephalogram (EEG)-based BCIs, particularly those that use steady-state visually evoked potentials (SSVEPs), have gained popularity in BCI research due to their non- intrusive nature, portability, and ease of setup [2]. This work digs into the improvements in SSVEP-based BCIs, specifically their neurophysiological responses that are coordinated with periodic visual stimuli [3]. The SSVEP BCI allocates unique frequencies to different targets, allowing for the detection of user intent via EEG signals when focusing on a certain target [4]. SSVEP BCIs are notable for their high information transfer rates (ITRs), low user training requirements, and the lack of individual decoder calibration. However, obstacles come from variances in visual stimulus properties that influence SSVEP responses, as well as the scarcity of commercial or clinical systems [5]. Notably, the SSVEP pattern outperforms other EEG signal patterns in terms of correct classification, making it a focus of current BCI research [6]. The primary goal of this study is to investigate the practical applications of SSVEP-based BCIs, namely speller systems. These