Comparing supervised and unsupervised approaches to emotion categorization in the human brain, body, and subjective experience.
Ontology highlight
ABSTRACT:
SUBMITTER: Azari B
PROVIDER: S-EPMC7679385 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:We investigated emotion classification from brief video recordings from the GEMEP database wherein actors portrayed 18 emotions. Vocal features consisted of acoustic parameters related to frequency, intensity, spectral distribution, and durations. Facial features consisted of facial action units. We first performed a series of person-independent supervised classification experiments. Best performance (AUC = 0.88) was obtained by merging the output from the best unimodal vocal (Elastic Net, AUC = 0.82) and facial (Random Forest, AUC = 0.80) classifiers using a late fusion approach and the product rule method. All 18 emotions were recognized with above-chance recall, although recognition rates varied widely across emotions (e.g., high for amusement, anger, and disgust; and low for shame). Multimodal feature patterns for each emotion are described in terms of the vocal and facial features that contributed most to classifier performance. Next, a series of exploratory unsupervised classification experiments were performed to gain more insight into how emotion expressions are organized. Solutions from traditional clustering techniques were interpreted using decision trees in order to explore which features underlie clustering. Another approach utilized various dimensionality reduction techniques paired with inspection of data visualizations. Unsupervised methods did not cluster stimuli in terms of emotion categories, but several explanatory patterns were observed. Some could be interpreted in terms of valence and arousal, but actor and gender specific aspects also contributed to clustering. Identifying explanatory patterns holds great potential as a meta-heuristic when unsupervised methods are used in complex classification tasks.
| S-EPMC8725659 | biostudies-literature
Project description:Humans can recognize emotions expressed through body motion with high accuracy even when the stimuli are impoverished. However, most of the research on body motion has relied on exaggerated displays of emotions. In this paper we present two experiments where we investigated whether emotional body expressions could be recognized when they were recorded during natural narration. Our actors were free to use their entire body, face, and voice to express emotions, but our resulting visual stimuli used only the upper body motion trajectories in the form of animated stick figures. Observers were asked to perform an emotion recognition task on short motion sequences using a large and balanced set of emotions (amusement, joy, pride, relief, surprise, anger, disgust, fear, sadness, shame, and neutral). Even with only upper body motion available, our results show recognition accuracy significantly above chance level and high consistency rates among observers. In our first experiment, that used more classic emotion induction setup, all emotions were well recognized. In the second study that employed narrations, four basic emotion categories (joy, anger, fear, and sadness), three non-basic emotion categories (amusement, pride, and shame) and the "neutral" category were recognized above chance. Interestingly, especially in the second experiment, observers showed a bias toward anger when recognizing the motion sequences for emotions. We discovered that similarities between motion sequences across the emotions along such properties as mean motion speed, number of peaks in the motion trajectory and mean motion span can explain a large percent of the variation in observers' responses. Overall, our results show that upper body motion is informative for emotion recognition in narrative scenarios.
| S-EPMC4075474 | biostudies-literature
Project description:Functional magnetic resonance imaging (fMRI) was used to examine cognitive regulation of negative emotion in 12 unmedicated patients with major depressive disorder (MDD) and 24 controls. The participants used reappraisal to increase (real condition) and reduce (photo condition) the personal relevance of negative and neutral pictures during fMRI as valence ratings were collected; passive viewing (look condition) served as a baseline. Reappraisal was not strongly affected by MDD. Ratings indicated that both groups successfully reappraised negative emotional experience. Both groups also showed better memory for negative vs. neutral pictures 2 weeks later. Across groups, increased brain activation was observed on negative/real vs. negative/look and negative/photo trials in left dorsolateral prefrontal cortex (DLPFC), rostral anterior cingulate, left parietal cortex, caudate, and right amygdala. Depressive severity was inversely correlated with activation modulation in the left DLPFC, right amygdala, and right cerebellum during negative reappraisal. The lack of group differences suggests that depressed adults can modulate the brain activation and subjective experience elicited by negative pictures when given clear instructions. However, the negative relationship between depression severity and effects of reappraisal on brain activation indicates that group differences may be detectable in larger samples of more severely depressed participants.
| S-EPMC3640775 | biostudies-literature
Project description:Music is often discussed to be emotional because it reflects expressive movements in audible form. Thus, a valid approach to measure musical emotion could be to assess movement stimulated by music. In two experiments we evaluated the discriminative power of mobile-device generated acceleration data produced by free movement during music listening for the prediction of ratings on the Geneva Emotion Music Scales (GEMS-9). The quality of prediction for different dimensions of GEMS varied between experiments for tenderness (R12(first experiment) = 0.50, R22(second experiment) = 0.39), nostalgia (R12 = 0.42, R22 = 0.30), wonder (R12 = 0.25, R22 = 0.34), sadness (R12 = 0.24, R22 = 0.35), peacefulness (R12 = 0.20, R22 = 0.35) and joy (R12 = 0.19, R22 = 0.33) and transcendence (R12 = 0.14, R22 = 0.00). For others like power (R12 = 0.42, R22 = 0.49) and tension (R12 = 0.28, R22 = 0.27) results could be almost reproduced. Furthermore, we extracted two principle components from GEMS ratings, one representing arousal and the other one valence of the experienced feeling. Both qualities, arousal and valence, could be predicted by acceleration data, indicating, that they provide information on the quantity and quality of experience. On the one hand, these findings show how music-evoked movement patterns relate to music-evoked feelings. On the other hand, they contribute to integrate findings from the field of embodied music cognition into music recommender systems.
| S-EPMC4945052 | biostudies-literature
Project description:Due to major artifacts and uncalibrated Hounsfield units (HU), cone-beam computed tomography (CBCT) cannot be used readily for diagnostics and therapy planning purposes. This study addresses image-to-image translation by convolutional neural networks (CNNs) to convert CBCT to CT-like scans, comparing supervised to unsupervised training techniques, exploiting a pelvic CT/CBCT publicly available dataset. Interestingly, quantitative results were in favor of supervised against unsupervised approach showing improvements in the HU accuracy (62% vs. 50%), structural similarity index (2.5% vs. 1.1%) and peak signal-to-noise ratio (15% vs. 8%). Qualitative results conversely showcased higher anatomical artifacts in the synthetic CBCT generated by the supervised techniques. This was motivated by the higher sensitivity of the supervised training technique to the pixel-wise correspondence contained in the loss function. The unsupervised technique does not require correspondence and mitigates this drawback as it combines adversarial, cycle consistency, and identity loss functions. Overall, two main impacts qualify the paper: (a) the feasibility of CNN to generate accurate synthetic CT from CBCT images, which is fast and easy to use compared to traditional techniques applied in clinics; (b) the proposal of guidelines to drive the selection of the better training technique, which can be shifted to more general image-to-image translation.
| S-EPMC8395013 | biostudies-literature
Project description:Inference of gene regulatory network from expression data is a challenging task. Many methods have been developed to this purpose but a comprehensive evaluation that covers unsupervised, semi-supervised and supervised methods, and provides guidelines for their practical application, is lacking. We performed an extensive evaluation of inference methods on simulated and experimental expression data. The results reveal low prediction accuracies for unsupervised techniques with the notable exception of the Z-SCORE method on knockout data. In all other cases, the supervised approach achieved the highest accuracies and even in a semi-supervised setting with small numbers of only positive samples, outperformed the unsupervised techniques.
| S-EPMC3956069 | biostudies-literature
Project description:This work aimed to identify pre-existing health conditions of patients with traumatic brain injury (TBI) and develop predictive models for the first TBI event and its external causes by employing a combination of unsupervised and supervised learning algorithms. We acquired up to five years of pre-injury diagnoses for 488,107 patients with TBI and 488,107 matched control patients who entered the emergency department or acute care hospitals between April 1st, 2002, and March 31st, 2020. Diagnoses were obtained from the Ontario Health Insurance Plan (OHIP) database which contains province-wide claims data by physicians in Ontario, Canada for inpatient and outpatient services. A screening process was conducted on the OHIP diagnostic codes to limit the subsequent analysis to codes that were predictive of TBI, which concluded that 314 codes were significantly associated with TBI. The Latent Dirichlet Allocation (LDA) model was applied to the diagnostic codes and generated an optimal number of 19 topics that concur with published literature but also suggest other unexplored areas. Estimated word-topic probabilities from the LDA model helped us detect pre-morbid conditions among patients with TBI by uncovering the underlying patterns of diagnoses, meanwhile estimated document-topic probabilities were utilized in variable creation as form of a dimension reduction. We created 19 topic scores for each patient in the cohort which were utilized along with socio-demographic factors for Random Forest binary classifier models. Test set performances evaluated using area under the receiver operating characteristic curve (AUC) were: TBI event (AUC = 0.85), external cause of injury: falls (AUC = 0.85), struck by/against (AUC = 0.83), cyclist collision (AUC = 0.76), motor vehicle collision (AUC = 0.83). Our analysis successfully demonstrated the feasibility of using machine learning to predict TBI due to various external causes and identified the most important factors that contribute to this prediction.
| S-EPMC10785655 | biostudies-literature
Project description:Emotion theorists have characterized emotions as involving coherent responding across various emotion response systems (e.g., covariation of subjective experience and physiology). Greater response system coherence has been theorized to promote well-being, yet very little research has tested this assumption. The current study examined whether individuals with greater coherence between physiology and subjective experience of emotion report greater well-being. We also examined factors that may predict the magnitude of coherence, such as emotion intensity, cognitive reappraisal, and expressive suppression. Participants (N = 63) completed self-report measures of well-being, expressive suppression, and cognitive reappraisal. They then watched a series of emotionally evocative film clips designed to elicit positive and negative emotion. During the films, participants continuously rated their emotional experience using a rating dial, and their autonomic physiological responses were recorded. Time-lagged cross-correlations were used to calculate within-participant coherence between intensity of emotional experience (ranging from neutral to very negative or very positive) and physiology (composite of cardiac interbeat interval, skin conductance, ear pulse transit time, finger pulse transit time and amplitude, systolic and diastolic blood pressure). Results indicated that individuals with greater coherence reported greater well-being. Coherence was highest during the most emotionally intense film and among individuals who reported lower expressive suppression. However, coherence was not associated with reappraisal. These findings provide support for the idea that greater emotion coherence promotes well-being and also shed light on factors that are associated with the magnitude of coherence. (PsycInfo Database Record (c) 2020 APA, all rights reserved).
| S-EPMC8158438 | biostudies-literature
Project description:Subjective cognitive decline (SCD) has been proposed as a risk factor for future cognitive decline and dementia. Given the heterogeneity of SCD and the lack of consensus about how to classify this condition, different operationalization approaches still need to be compared. In this study, we used the same sample of individuals to compare different SCD operationalization approaches. We included 399 cognitively healthy individuals from a community-based cohort. SCD was assessed through nine questions about memory and non-memory subjective complaints. We applied four approaches to operationalize SCD: two hypothesis-driven approaches and two data-driven approaches. We characterized the resulting groups from each operationalization approach using multivariate methods on comprehensive demographic, clinical, cognitive, and neuroimaging data. We identified two main phenotypes: an amnestic phenotype characterized by an Alzheimer's Disease (AD) signature pattern of brain atrophy; and an anomic phenotype, which was mainly related to cerebrovascular pathology. Furthermore, language complaints other than naming helped to identify a subgroup with subclinical cognitive impairment and difficulties in activities of daily living. This subgroup also showed an AD signature pattern of atrophy. The identification of SCD phenotypes, characterized by different syndromic and biomarker profiles, varies depending on the operationalization approach used. In this study we discuss how these findings may be used in clinical practice and research.
| S-EPMC7902653 | biostudies-literature
Project description:The nature of the code used in the auditory cortex to represent complex auditory stimuli, such as naturally spoken words, remains a matter of debate. Here we argue that such representations are encoded by stable spatio-temporal patterns of firing within cell assemblies known as polychronous groups, or PGs. We develop a physiologically grounded, unsupervised spiking neural network model of the auditory brain with local, biologically realistic, spike-time dependent plasticity (STDP) learning, and show that the plastic cortical layers of the network develop PGs which convey substantially more information about the speaker independent identity of two naturally spoken word stimuli than does rate encoding that ignores the precise spike timings. We furthermore demonstrate that such informative PGs can only develop if the input spatio-temporal spike patterns to the plastic cortical areas of the model are relatively stable.
| S-EPMC5552261 | biostudies-literature