Journal of Intelligent & Fuzzy Systems - Volume 45, issue 5

Authors: Fu, Pengbin | Ma, Yuchen | Yang, Huirong

Article Type: Research Article

Abstract: The speaker diarization task pertains to the automated differentiation of speakers within an audio recording, while lacking any prior information regarding the speakers. The introduction of the self-attention mechanism in End-to-End Neural Speaker Diarization (EEND) has elegantly resolved the issue of overlapping speakers. The Transformer model equipped with self-attention mechanism has shown great potential in collecting global information, yielding remarkable outcomes in various tasks. However, the individual speaker characteristics are predominantly reflected in the contextual information, which conventional self-attention would not adequately address. In this study, we propose a hierarchical encoders model to augment the encoders’ acquisition of speaker information …in two distinct ways: (1) Constraining the perceptual field of the self-attentive mechanism with left-right windows or Gaussian weights to highlight contextual information; (2) Utilizing a pre-trained time-delay neural network based speaker embedding extractor to alleviate the shortcomings of speaker feature extraction ability. We evaluate the proposed methods on a simulated dataset of two speakers and a real conversation dataset. The model with the most favorable outcomes among the proposed enhancements achieves a diarization error rate of 7.74% on the simulated dataset and 21.92% on MagicData-RAMC after adaptation. These results compellingly demonstrate the efficacy of the proposed methods. Show more

Keywords: Speaker diarization, contextual information, Gaussian weight, constraint self-attention

DOI: 10.3233/JIFS-230249

Citation: Journal of Intelligent & Fuzzy Systems, vol. 45, no. 5, pp. 9169-9180, 2023

Authors: Qin, Ying

Article Type: Research Article

Abstract: English language teaching varies with the universities and faculties for improving student knowledge through adaptability. In improving the adaptability features, multiple practices are blended based on previous outcomes. The outcomes are considered through the accumulated big data for leveraging student performance. This article introduces a Blended Model using Big Data Analytics (BM-BDA) to provide an upgraded teaching environment for different students. This study applied learning analytics and educational big data methods for the early prediction of students’ final academic performance in a blended model for English teaching. The model aims at rectifying the performance inaccuracies observed in the previous sessions …through the pursued teaching methods. Furthermore, the identification is pursued using teaching model classification and its results over students’ performance. The classification is pursued using conventional classifier learning based on different inaccuracies. The inaccuracy in teaching efficiency using the implied model is classified for different types of students for step-by-step model tuning. The tuning is performed by inheriting the successful implications from the other methods. This improves the inclusion and blending of the diverse method to a required level for teaching efficiency. The successful blending method is discarded from the classification process post the outcome verification. This requires intense data analysis using diverse student performance and implied teaching methods. Show more

Keywords: Big data, blended models, classification learning, English teaching

DOI: 10.3233/JIFS-230842

Citation: Journal of Intelligent & Fuzzy Systems, vol. 45, no. 5, pp. 9181-9197, 2023