Content uploaded by Paul Nussbaum
Author content
All content in this area was uploaded by Paul Nussbaum on May 02, 2019
Content may be subject to copyright.
A preview of the PDF is not available
Signal Processing of Electroencephalogram for the Detection of Attentiveness towards Short Training Videos
Abstract and Figures
This research has developed a novel method which uses an easy to deploy single dry electrode wireless electroencephalogram (EEG) collection device as an input to an automated system that measures indicators of a participant's attentiveness while they are watching a short training video. The results are promising, including 85% or better accuracy in identifying whether a participant is watching a segment of video from a boring scene or lecture, versus a segment of video from an attentiveness inducing active lesson or memory quiz. In addition, the final system produces an ensemble average of attentiveness across many participants, pinpointing areas in the training videos that induce peak attentiveness. Qualitative analysis of the results of this research is also very promising. The system produces attentiveness graphs for individual participants and these triangulate well with the thoughts and feelings those participants had during different parts of the videos, as described in their own words.
As distance learning and computer based training become more popular, it is of great interest to measure if students are attentive to recorded lessons and short training videos. This research was motivated by this interest, as well as recent advances in electronic and computer engineering's use of biometric signal analysis for the detection of affective (emotional) response. Signal processing of EEG has proven useful in measuring alertness, emotional state, and even towards very specific applications such as whether or not participants will recall television commercials days after they have seen them. This research extended these advances by creating an automated system which measures attentiveness towards short training videos.
The bulk of the research was focused on electrical and computer engineering, specifically the optimization of signal processing algorithms for this particular application. A review of existing methods of EEG signal processing and feature extraction methods shows that there is a common subdivision of the steps that are used in different EEG applications. These steps include hardware sensing filtering and digitizing, noise removal, chopping the continuous EEG data into windows for processing, normalization, transformation to extract frequency or scale information, treatment of phase or shift information, and additional post-transformation noise reduction techniques. A large degree of variation exists in most of these steps within the currently documented state of the art. This research connected these varied methods into a single holistic model that allows for comparison and selection of optimal algorithms for this application.
The research described herein provided for such a structured and orderly comparison of individual signal analysis and feature extraction methods. This study created a concise algorithmic approach in examining all the aforementioned steps. In doing so, the study provided the framework for a systematic approach which followed a rigorous participant cross validation so that options could be tested, compared and optimized. Novel signal analysis methods were also developed, using new techniques to choose parameters, which greatly improved performance.
The research also utilizes machine learning to automatically categorize extracted features into measures of attentiveness. The research improved existing machine learning with novel methods, including a method of using per-participant baselines with kNN machine learning. This provided an optimal solution to extend current EEG signal analysis methods that were used in other applications, and refined them for use in the measurement of attentiveness towards short training videos.
These algorithms are proven to be best via selection of optimal signal analysis and optimal machine learning steps identified through both n-fold and participant cross validation. The creation of this new system which uses signal processing of EEG for the detection of attentiveness towards short training videos has created a significant advance in the field of attentiveness measuring towards short training videos.
http://search.proquest.com/docview/1500840077
Figures - uploaded by Paul Nussbaum
Author content
All figure content in this area was uploaded by Paul Nussbaum
Content may be subject to copyright.
Supplementary resources (2)
ResearchGate has not been able to resolve any citations for this publication.
This paper investigates how the use of machine learning techniques can significantly predict the three major dimensions of learner's emotions (pleasure, arousal and dominance) from brainwaves. This study has adopted an experimentation in which participants were exposed to a set of pictures from the International Affective Picture System (IAPS) while their electrical brain activity was recorded with an electroencephalogram (EEG). The pictures were already rated in a previous study via the affective rating system Self-Assessment Manikin (SAM) to assess the three dimensions of pleasure, arousal, and dominance. For each picture, we took the mean of these values for all subjects used in this previous study and associated them to the recorded brainwaves of the participants in our study. Correlation and regression analyses confirmed the hypothesis that brainwave measures could significantly predict emotional dimensions. This can be very useful in the case of impassive, taciturn or disabled learners. Standard classification techniques were used to assess the reliability of the automatic detection of learners' three major dimensions from the brainwaves. We discuss the results and the pertinence of such a method to assess learner's emotions and integrate it into a brainwave- sensing Intelligent Tutoring System.
Recent progress in cognitive neuroscience suggests that alpha activity may reflect selective cortical inhibition involved in signal amplification, rather than neural idling. Unfortunately, these theoretical advances remain largely ignored in affective neuroscience. To address this limitation the present paper proposes a novel research avenue aimed at using alpha to elucidate cortical inhibitory mechanisms involved in affective processes. The proposal is illustrated by developing inhibitory accounts of affective attention and affective tuning phenomena. The emergent predictions were tested using event-related perturbations from 73 students evaluating affective and nonaffective aspects of five types of emotional images. The results revealed that upper alpha power was increased by affective content in general and aversive stimuli in particular from 350 ms at posterior and from 575 ms at central sites. The evaluation task interacted with affective content only at a liberal statistical significance level in late posterior alpha. These results are generally in line with the proposed inhibitory accounts of affective attention and tuning, although the evidence is preliminary rather than conclusive. As confirmation of functional origins of alpha in affect remains beyond the scope of a single study, this paper aims to inspire further extrapolation of the inhibitory account of alpha within affective neuroscience.
In recent studies of the structure of affect, positive and negative affect have consistently emerged as two dominant and relatively independent dimensions. A number of mood scales have been created to measure these factors; however, many existing measures are inadequate, showing low reliability or poor convergent or discriminant validity. To fill the need for reliable and valid Positive Affect and Negative Affect scales that are also brief and easy to administer, we developed two 10-item mood scales that comprise the Positive and Negative Affect Schedule (PANAS). The scales are shown to be highly internally consistent, largely uncorrelated, and stable at appropriate levels over a 2-month time period. Normative data and factorial and external evidence of convergent and discriminant validity for the scales are also presented. (PsycINFO Database Record (c) 2010 APA, all rights reserved)
The purpose of this article was to provide a comprehensive review of research on video podcasts from 2002 to 2011 in order to guide future studies and educational practice. Fifty-three, peer-reviewed articles were selected from an extensive search of the literature. Key topics included the history and growth of video podcasts, types of podcasts, previous literature reviews, benefits and challenges of using video podcasts, methodological concerns, and suggestions for future research. Key benefits included positive affective and cognitive attitudes toward video podcasts, control over learning, improved study habits, and increased learning performance. Key challenges included a variety of technical problems, preference of some students for lectures, and reduced class attendance. Methodological concerns involved insufficient description of video podcasts examined, limited sample selection and description, and the absence of reliability and validity estimates for data collection tools. Suggestions for future research include focusing on the quality and design of video podcasts, pedagogical strategies, viewing patterns and impact on learning effectiveness, and in individual differences in video podcast use.
We discuss the strong relationship between affect and cognition and the importance of emotions in multimodal human computer interaction (HCI) and user modeling. We introduce the overall paradigm for our multimodal system that aims at recognizing its users' emotions and at responding to them accordingly depending upon the current context or application. We then describe the design of the emotion elicitation experiment we conducted by collecting, via wearable computers, physiological signals from the autonomic nervous system (galvanic skin response, heart rate, temperature) and mapping them to certain emotions (sadness, anger, fear, surprise, frustration, and amusement). We show the results of three different supervised learning algorithms that categorize these collected signals in terms of emotions, and generalize their learning to recognize emotions from new collections of signals. We finally discuss possible broader impact and potential applications of emotion recognition for multimodal intelligent systems.
Electroencephalographic (EEG) recordings are often contaminated with muscle artifacts. This disturbing myogenic activity not only strongly affects the visual analysis of EEG, but also most surely impairs the results of EEG signal processing tools such as source localization. This article focuses on the particular context of the contamination epileptic signals (interictal spikes) by muscle artifact, as EEG is a key diagnosis tool for this pathology. In this context, our aim was to compare the ability of two stochastic approaches of blind source separation, namely independent component analysis (ICA) and canonical correlation analysis (CCA), and of two deterministic approaches namely empirical mode decomposition (EMD) and wavelet transform (WT) to remove muscle artifacts from EEG signals. To quantitatively compare the performance of these four algorithms, epileptic spike-like EEG signals were simulated from two different source configurations and artificially contaminated with different levels of real EEG-recorded myogenic activity. The efficiency of CCA, ICA, EMD, and WT to correct the muscular artifact was evaluated both by calculating the normalized mean-squared error between denoised and original signals and by comparing the results of source localization obtained from artifact-free as well as noisy signals, before and after artifact correction. Tests on real data recorded in an epileptic patient are also presented. The results obtained in the context of simulations and real data show that EMD outperformed the three other algorithms for the denoising of data highly contaminated by muscular activity. For less noisy data, and when spikes arose from a single cortical source, the myogenic artifact was best corrected with CCA and ICA. Otherwise when spikes originated from two distinct sources, either EMD or ICA offered the most reliable denoising result for highly noisy data, while WT offered the better denoising result for less noisy data. These results suggest that the performance of muscle artifact correction methods strongly depend on the level of data contamination, and of the source configuration underlying EEG signals. Eventually, some insights into the numerical complexity of these four algorithms are given.
Optimum use of electroencephalography (EEG) can be improved by an understanding of the underlying signals and the nature of the recording and display technology. Topics reviewed in this section include the source of EEG signals; electrical noise; behavior of the amplifiers, filters, sampling, and displays; construction of EEG montages; quantitative EEG; and electrical safety. The author discusses how these considerations impact data quality and interpretation.
This paper presents our preliminary study EEG brain signals of children with attention deficit hyperactivity disorder (ADHD) in order to support a computer assisted diagnostic system. The EEG signals were recorded from 13 children including normal and children diagnosed with ADHD. We analyzed the signals with multilevel discrete wavelet decompositions in order to extract brain signal power spectrum features. A wavelet thresholding technique was employed to further improve the data quality by denoising the artifacts. In order to discriminate the attention level in electrical brain activity of ADHD children, we used a standard Self-Organizing Map clustering technique with wavelet coefficient input features. Clustering results varied depending on the wavelet feature extraction stage, particularly it was noticed that accuracy was dependent on the type of the used wavelet function. The clustering results demonstrate that `sym7' wavelet function provides better input feature localization to provide the accurate separation of normal and disordered children's brain activity.
We propose a model to explain the dynamics of affective states that emerge during deep learning activities. The model predicts that learners in a state of engagement/flow will experience cognitive disequilibrium and confusion when they face contradictions, incongruities, anomalies, obstacles to goals, and other impasses. Learners revert into the engaged/flow state if equilibrium is restored through thought, reflection, and problem solving. However, failure to restore equilibrium as well as obstacles that block goals trigger frustration, which, if unresolved, will eventually lead to boredom. The major hypotheses of the model were supported in two studies in which participants completed a 32–35min tutoring session with a computer tutor. Their affective states were tracked at approximately 110 points in their tutoring sessions via a retrospective affect judgment protocol. Time series analyses confirmed the presence of confusion–engagement/flow, boredom–frustration, and confusion–frustration oscillations. We discuss enhancements of the model to address individual differences and pedagogical and motivational strategies that are inspired by the model.
The usability analysis of information systems has been the target of several research studies over the past thirty years. These studies have highlighted a great diversity of points of view, including researchers from different scientific areas such as Ergonomics, Computer Science, Design and Education. Within the domain of information ergonomics, the study of tools and methods used for usability evaluation dedicated to E-learning presents evidence that there is a continuous and dynamic evolution of E-learning systems, in many different contexts -academics and corporative. These systems, also known as LMS (Learning Management Systems), can be classified according to their educational goals and their technological features. However, in these systems the usability issues are related with the relationship/interactions between user and system in the user's context. This review is a synthesis of research project about Information Ergonomics and embraces three dimensions, namely the methods, models and frameworks that have been applied to evaluate LMS. The study also includes the main usability criteria and heuristics used. The obtained results show a notorious change in the paradigms of usability, with which it will be possible to discuss about the studies carried out by different researchers that were focused on usability ergonomic principles aimed at E-learning.