Subtle Expressions Research Articles

Radiation synthesis of high conductivity hydrogel based on tragacanth gum/poly (ionic liquids) for multimodal sensors and supercapacitor

Natural polymer-based hydrogels have found extensive use in flexible sensing, energy storage, and other fields because of their environmental sustainability and biocompatibility. Nonetheless, numerous challenges persist in the development of hydrogels with outstanding conductivity solely from natural polymers. Herein, we have successfully synthesized hydrogels based on natural polymer (tragacanth gum) and ionic liquids (1-vinyl-3-ethylimidazolium bromide) using a convenient and efficient one-step ionizing radiation method (TG/PIL hydrogels). The TG/PIL hydrogels exhibit high ionic conductivity (7.1 S m−1 at 25 °C), and can be used for multimodal sensors, including strain and temperature sensors. It has exceptional capabilities in monitoring human motor behavior, capturing subtle facial expressions and pulses beat. TG/PIL hydrogel can also accurately sense changes in the temperature of the external environment, and have significant thermal sensitivity within the range of 40 to 60 °C (−3.22 % /°C). Furthermore, the high conductivity of TG/PIL hydrogels enables them to exhibit outstanding performance in supercapacitor electrolytes, it has good stability in a certain load bearing range, temperature range, folding angle range. This work offers a straightforward technique for creating a multimodal hydrogel sensor, with promising applications in flexible wearable devices, energy storage, and beyond.

Deep learning detects subtle facial expressions in a multilevel society primate.

Facial expressions in nonhuman primates are complex processes involving psychological, emotional, and physiological factors, and may use subtle signals to communicate significant information. However, uncertainty surrounds the functional significance of subtle facial expressions in animals. Using artificial intelligence (AI), this study found that nonhuman primates exhibit subtle facial expressions that are undetectable by human observers. We focused on the golden snub-nosed monkeys (Rhinopithecus roxellana), a primate species with a multilevel society. We collected 3427 front-facing images of monkeys from 275 video clips captured in both wild and laboratory settings. Three deep learning models, EfficientNet, RepMLP, and Tokens-To-Token ViT, were utilized for AI recognition. To compare the accuracy of human performance, two groups were recruited: one with prior animal observation experience and one without any such experience. The results showed human observers to correctly detect facial expressions (32.1% for inexperienced humans and 45.0% for experienced humans on average with a chance level of 33%). In contrast, the AI deep learning models achieved significantly higher accuracy rates. The best-performing model achieved an accuracy of 94.5%. Our results provide evidence that golden snub-nosed monkeys exhibit subtle facial expressions. The results further our understanding of animal facial expressions and also how such modes of communication may contribute to the origin of complex primate social systems.

A Multi-Class Facial Emotion Recognition System for Still Images 1Omage Micheal, 2Fasola Olusanjo & 3Woods Nancy C. 1,3University of Ibadan, Ibadan, Nigeria 2Federal University of Technology, Minna, Nigeria E-mails: 1micheal.omage@gmail.com, 2Sanjo.fasola@futminna.edu.ng, 3Chyn.woods@gmail.com Phone Nos: 1+2348084121913, 2+2348023138034, 3+2348037273291 ABSTRACT Facial expressions serve as a potent and natural means of human communication, conveying a wide range of emotions. In the realm of

Facial expressions serve as a potent and natural means of human communication, conveying a wide range of emotions. In the realm of artificial intelligence and computer vision, recognizing these emotions from facial cues has gained significant importance. Most research focus on detecting a single dominant emotion on any given facial image, although an image could portray several emotions. This work presents a multi-class facial emotion recognition system capable of predicting and quantifying the presence of multiple emotions in a single facial image. The system employs a Convolutional Neural Network (CNN) architecture optimized for multi-label classification across eight emotion classes. The AffectNet dataset, comprising 27,836 images with diverse ages, ethnicities, and real-world settings, was utilized for training the model, to enhance generalization capabilities. The model's performance was evaluated on an impartial benchmark dataset, FER-Plus, containing 35,710 images. The CNN model performed well, with a micro F1-score of 0.7 and a macro F1-score of 0.7 on the test set from AffectNet, and a micro F1-score of 0.7 and a macro F1-score of 0.4 on the FER-Plus dataset. Qualitative analysis demonstrated the model's capability in recognizing blended and subtle facial expressions, providing significant probability distributions across multiple emotion labels. The developed system advances the state-of-the-art in multi-label facial emotion recognition by successfully modeling emotion mixtures, contributing to a deeper understanding of this challenging task and enabling practical applications in affective computing, human-computer interaction, psychological studies, and improving accessibility and inclusivity in emotional computing technologies. Keywords: Multi-Class, Facial Emotion Recognition, Security, System, Still Images, Models, Detection Omage, M. Fasola, O. & Woods, N.C. (2024): A Multi-Class Facial Emotion Recognition System for Still Images. Journal of Advances in Mathematical & Computational Science. Vol. 12, No. 3. Pp 67-80. Available online at www.isteams.net/mathematics-computationaljournal. dx.doi.org/10.22624/AIMS/MATHS/V12N3P6

A real-world multi-center RNA-seq benchmarking study using the Quartet and MAQC reference materials

Translating RNA-seq into clinical diagnostics requires ensuring the reliability and cross-laboratory consistency of detecting clinically relevant subtle differential expressions, such as those between different disease subtypes or stages. As part of the Quartet project, we present an RNA-seq benchmarking study across 45 laboratories using the Quartet and MAQC reference samples spiked with ERCC controls. Based on multiple types of ‘ground truth’, we systematically assess the real-world RNA-seq performance and investigate the influencing factors involved in 26 experimental processes and 140 bioinformatics pipelines. Here we show greater inter-laboratory variations in detecting subtle differential expressions among the Quartet samples. Experimental factors including mRNA enrichment and strandedness, and each bioinformatics step, emerge as primary sources of variations in gene expression. We underscore the profound influence of experimental execution, and provide best practice recommendations for experimental designs, strategies for filtering low-expression genes, and the optimal gene annotation and analysis pipelines. In summary, this study lays the foundation for developing and quality control of RNA-seq for clinical diagnostic purposes.

Universal Facial Encoding of Codec Avatars from VR Headsets

Faithful real-time facial animation is essential for avatar-mediated telepresence in Virtual Reality (VR). To emulate authentic communication, avatar animation needs to be efficient and accurate: able to capture both extreme and subtle expressions within a few milliseconds to sustain the rhythm of natural conversations. The oblique and incomplete views of the face, variability in the donning of headsets, and illumination variation due to the environment are some of the unique challenges in generalization to unseen faces. In this paper, we present a method that can animate a photorealistic avatar in realtime from head-mounted cameras (HMCs) on a consumer VR headset. We present a self-supervised learning approach, based on a cross-view reconstruction objective, that enables generalization to unseen users. We present a lightweight expression calibration mechanism that increases accuracy with minimal additional cost to run-time efficiency. We present an improved parameterization for precise ground-truth generation that provides robustness to environmental variation. The resulting system produces accurate facial animation for unseen users wearing VR headsets in realtime. We compare our approach to prior face-encoding methods demonstrating significant improvements in both quantitative metrics and qualitative results.

Potentially Queer: Mass Cultural Images in Vanity Fair and George Platt Lynes’s Scrapbooks

ABSTRACT This article examines George Platt Lynes’s scrapbooks, highlighting their role in the queer cultural narrative of the 1930s. Lynes, a key figure in New York’s queer artistic circle, blended personal and mass media images in his scrapbooks, including Vanity Fair clippings. These collections reflect the era’s subtle queer expressions and provide unique insights into queer history and media. Focusing on a Cary Grant photograph from Vanity Fair, the article discusses the multifaceted interpretations of visual media in understanding queer narratives within modernism and modernity and emphasizes the integral role of queerness within these frameworks.

Collegiality and Dissent in Polish Administrative Courts

This article addresses a gap in existing research by focusing on the often-neglected realm of judicial interactions and internal dynamics within specific courts concerning the phenomenon of votum separatum. We examine the forms and practices of collegiality within Polish administrative courts and their influence on judges' decisions to file dissenting opinions. Additionally, we investigate the reactions of fellow judges when a dissent is announced. Our qualitative research methodology relies on in-depth interviews to prevent the imposition of predefined categories. Participants were encouraged to recount their experiences related to composing or participating in decisions involving dissenting opinions. This approach led to the emergence of categories related to collegiality, its functions, and inherent tensions. Our findings reveal that collegiality manifests in various forms beyond panel deliberations. Notably, our research uncovers the existence of departmental meetings in provincial administrative courts where issues addressed in dissenting opinions are discussed. Furthermore, judges' perspectives indicate that the most common scenario leading to dissenting opinions arises when judges from different panels reach opposing decisions. This dilemma prompts judges to choose between adhering to the initial panel's decision or voting for a divergent position proposed by the second panel. Finally, our observations within courtrooms highlight that the ideal of the dispassionate judge does not exclude subtle expressions of surprise or disappointment. These findings enrich our understanding of judicial interactions, shedding light on the complexities of collegiality and dissent within the context of Polish administrative courts.

Multimodal Sensing for Depression Risk Detection: Integrating Audio, Video, and Text Data.

Depression is a major psychological disorder with a growing impact worldwide. Traditional methods for detecting the risk of depression, predominantly reliant on psychiatric evaluations and self-assessment questionnaires, are often criticized for their inefficiency and lack of objectivity. Advancements in deep learning have paved the way for innovations in depression risk detection methods that fuse multimodal data. This paper introduces a novel framework, the Audio, Video, and Text Fusion-Three Branch Network (AVTF-TBN), designed to amalgamate auditory, visual, and textual cues for a comprehensive analysis of depression risk. Our approach encompasses three dedicated branches-Audio Branch, Video Branch, and Text Branch-each responsible for extracting salient features from the corresponding modality. These features are subsequently fused through a multimodal fusion (MMF) module, yielding a robust feature vector that feeds into a predictive modeling layer. To further our research, we devised an emotion elicitation paradigm based on two distinct tasks-reading and interviewing-implemented to gather a rich, sensor-based depression risk detection dataset. The sensory equipment, such as cameras, captures subtle facial expressions and vocal characteristics essential for our analysis. The research thoroughly investigates the data generated by varying emotional stimuli and evaluates the contribution of different tasks to emotion evocation. During the experiment, the AVTF-TBN model has the best performance when the data from the two tasks are simultaneously used for detection, where the F1 Score is 0.78, Precision is 0.76, and Recall is 0.81. Our experimental results confirm the validity of the paradigm and demonstrate the efficacy of the AVTF-TBN model in detecting depression risk, showcasing the crucial role of sensor-based data in mental health detection.

Optimal processing of surface facial EMG to identify emotional expressions: A data-driven approach

Surface facial electromyography (EMG) is commonly used to detect emotions from subtle facial expressions. Although there are established procedures for collecting EMG data and some aspects of their processing, there is little agreement among researchers about the optimal way to process the EMG signal, so that the study-unrelated variability (noise) is removed, and the emotion-related variability is best detected. The aim of the current paper was to establish an optimal processing pipeline for EMG data for identifying emotional expressions in facial muscles. We identified the most common processing steps from existing literature and created 72 processing pipelines that represented all the different processing choices. We applied these pipelines to a previously published dataset from a facial mimicry experiment, where 100 adult participants observed happy and sad facial expressions, whilst the activity of their facial muscles, zygomaticus major and corrugator supercilii, was recorded with EMG. We used a resampling approach and subsets of the original data to investigate the effect and robustness of different processing choices on the performance of a logistic regression model that predicted the mimicked emotion (happy/sad) from the EMG signal. In addition, we used a random forest model to identify the most important processing steps for the sensitivity of the logistic regression model. Three processing steps were found to be most impactful: baseline correction, standardisation within muscles, and standardisation within subjects. The chosen feature of interest and the signal averaging had little influence on the sensitivity to the effect. We recommend an optimal processing pipeline, share our code and data, and provide a step-by-step walkthrough for researchers.

Mu-Suppression Neurofeedback Training Targeting the Mirror Neuron System: A Pilot Study

Neurofeedback training (NFT) is a promising adjuvant intervention method. The desynchronization of mu rhythm (8–13 Hz) in the electroencephalogram (EEG) over centro-parietal areas is known as a valid indicator of mirror neuron system (MNS) activation, which has been associated with social skills. Still, the effect of neurofeedback training on the MNS requires to be well investigated. The present study examined the possible impact of NFT with a mu suppression training protocol encompassing 15 NFT sessions (45 min each) on 16 healthy neurotypical participants. In separate pre- and post-training sessions, 64-channel EEG was recorded while participants (1) observed videos with various types of movements (including complex goal-directed hand movements and social interaction scenes) and (2) performed the "Reading the Mind in the Eyes Test" (RMET). EEG source reconstruction analysis revealed statistically significant mu suppression during hand movement observation across MNS-attributed fronto-parietal areas after NFT. The frequency analysis showed no significant mu suppression after NFT, despite the fact that numerical mu suppression appeared to be visible in a majority of participants during goal-directed hand movement observation. At the behavioral level, RMET accuracy scores did not suggest an effect of NFT on the ability to interpret subtle emotional expressions, although RMET response times were reduced after NFT. In conclusion, the present study exhibited preliminary and partial evidence that mu suppression NFT can induce mu suppression in MNS-attributed areas. More powerful experimental designs and longer training may be necessary to induce substantial and consistent mu suppression, particularly while observing social scenarios.

Beyond Words

Beyond Words

Discriminatory, yet socially accepted? Targets’ perceptions of subtle and blatant expressions of ethno-racial prejudice

IntroductionExtant research has studied prejudice expressions mainly from the majority perspective. We examined whether prejudice expressions conceptualized as subtle (vs. blatant) are perceived differently by their targets.MethodsParticipants who self-identified as potential targets of ethno-racial or religious prejudice (e.g., anti-Muslim, anti-Asian, anti-Arab, anti-Slavic, and anti-Black prejudice), answered questions about verbal expressions of prejudice taken from the subtle and blatant prejudice scales in an online survey.ResultsItems of the subtle (vs. blatant) prejudice scale were rated as more socially accepted and less discriminatory, elicited less negative emotions, and were reported to be experienced more often. Subtle expressions of prejudice were not more familiar to participants than blatant ones. Remarkably, blatant prejudice expressions were also perceived as relatively socially accepted and subtle prejudices as relatively discriminatory, as indicated by mean ratings above the scale midpoint. Lower discrimination ratings of subtle (vs. blatant) prejudice expressions were mainly due to perceptions of expressions exaggerating cultural differences. Exploratory analyses indicate that participants who reported more (vs. less) frequent overall exposure to the prejudice expressions perceived subtle and blatant stimuli as similarly discriminatory. This finding is compatible with the idea that individuals factor their personal experiences with prejudice into their assessments of discrimination.DiscussionWe discuss implications for interventions, especially regarding the social acceptability of blatant prejudice and promoting awareness of the discriminatory impact of subtle prejudice.

237 SeeMe: An Automated Methodology to Detect Early Signs of Recovery After Acute Brain Injury

INTRODUCTION: Assessing the functional state of the brain and predicting recovery after acute brain injury (ABI) remains a major unsolved problem. At present, physicians make predictions using clinical examination, traditional scoring systems, and statistical models, which have significant shortcomings and lack the ability to forecast recovery on an individual basis. METHODS: SeeMe is based on analyzing subtle facial expressions in response to auditory commands (i.e., “open your eyes,” “make a smile”) over time and compares the results to a pre-stimulus baseline. These subtle motions are often not detectable by the naked eye. We enrolled 26 comatose ABI patients and 16 healthy volunteers to test this tool. Command-evoked eye-opening movements were compared to clinical observation and the JFK Coma Recovery Scale-Revised (CRS-R). RESULTS: In our study, 22 of 26 (85%) patients recovered consciousness during hospitalization after ABI. On average, SeeMe detected command-evoked eye opening ∼9 days before clinical assessment using the CRS-R and ∼3 days before blinded human raters who observed the recorded videos using the naked eye.Using the optimally identified decision threshold, SeeMe demonstrated the following performance metrics on ABI patients with no eye opening to command (auditory CRS subscale = 2): sensitivity = 74.1%, specificity = 66.6%, positive predictive value = 16.8%, and negative predictive value = 96.6%. CONCLUSIONS: These findings demonstrate the high predictive performance of SeeMe in assessing consciousness in ABI patients and provide a quantitative tool for studying neural circuits and developing novel therapeutic approaches. SeeMe's ability to detect low-amplitude motor movements opens new avenues for understanding and facilitating recovery of consciousness in patients with ABI.

Microaggressions towards People with Mental Illness

IntroductionMicroaggressions, or subtle expressions of discrimination directed towards individuals because of their membership in marginalized social groups, are the subject of a growing body of literature (Sue, 2010). As a result of growing understanding of politically correct beliefs over time, they’ve been defined as subtler types of discrimination that have replaced formerly overt discrimination. Microaggressions differ from traditional prejudice in that they are frequently perpetrated by well-intentioned people who are oblivious of the negative implications and consequences of their conduct. Microaggressions have been documented in a variety of social groups, including racial/ethnic minorities (Sue et al., 2008; Torres et al., 2010), gender (Swim et al., 2001), sexual orientation (Shelton and Delgado-Romero, 2011), and ability status (Shelton and Delgado-Romero, 2011). Many people with mental illnesses have reported social rejection experiences that are similar to microaggressions, according to research (Cechnicki et al., 2011; Lundberg et al., 2009; Wright et al., 2000; Yanos et al., 2001).ObjectivesExisting measures of stigmatizing attitudes and behaviors may not capture much of the nuance in behavior that people with mental illness report to be particularly upsetting, so we thought it would be important to examine reliability and validity of the mental illness microaggressions scale-perpetrator version (MIMS-P) for measuring microaggression behavior in the general public in Turkey.Methods The methodological study will be conducted to establish the validity and reliability of the The mental illness microaggressions scale-perpetrator version (MIMS-P) scale to Turkish Culture and to determine the microaggression levels against individuals with mental illness in the general population. The sample of the study will consist of individuals who are reached through an online questionnaire and who agree to participate in the study. Individuals who have psychiatric disorders will not be included in the study.ResultsData collection process is still ongoing. Description of studies and the key findings will be presented.ConclusionsThe MIMS-P is designed to aid future study on the frequency of endorsement of microaggressions performed against people with mental illnesses, with the ultimate goal of understanding the mechanisms that lead to these acts.The development of an extra scale to measure microaggressions from the perspective of people with mental illnesses who encounter them is one of the future research objectives.With a better knowledge of these viewpoints and how they interact, effective therapies and public policy initiatives for reducing stigma against mental illness can be developed.Disclosure of InterestNone Declared

Integrating gender expertise into the Canadian Armed Forces: challenges for change agents and culture change.

Gender Advisors (GENADs) have played a key role in the efforts of military organizations worldwide to integrate gender perspectives, and culture change, within the defence and security context. Military organizations, however, continue to face challenges in regard to diversity and inclusion, including limited representation of women and other diverse groups who do not fit the white male, masculine stereotype, and subtle and overt expressions of prejudice and stigma towards under-represented and marginalized groups. In such an organizational context, the integration of gender perspectives has faced challenges, and transformative culture change has remained elusive. In particular, the experience of GENADs suggests that there may be unique challenges to serving as "gender experts" within military organizations. This paper, therefore, examines the lived experience of GENADs within the context of military organizations, as illustrated by GENADs in the Canadian Armed Forces. We consider two qualitative studies on the lived experience of GENADs and focus on the shared theme of legitimacy of gender expertise at both individual and systemic levels. This analysis highlights challenges that gendered power relations may pose for GENADs as individual change agents, and for systemic, transformative culture change, within existing military organizations, while reaffirming the importance of understanding the lived experience of GENADs in their pursuit of more equitable institutional and operational outcomes. Using social-psychological theories of tokenism, we consider more broadly what it means to be the gender person within masculinized military organizations and conclude with reflections on the potential contours of transformative culture change within the military context.

Borderline decisions?: Lack of justification for automatic deception detection at EU borders

Between 2016 and 2019, the European Union funded the development and testing of a system called “iBorderCtrl”, which aims to help detect illegal migration. Part of iBorderCtrl is an automatic deception detection system (ADDS): Using artificial intelligence, ADDS is designed to calculate the probability of deception by analyzing subtle facial expressions to support the decision-making of border guards. This text explains the operating principle of ADDS and its theoretical foundations. Against this background, possible deficits in the justification of the use of this system are pointed out. Finally, based on empirical findings, potential societal ramifications of an unjustified use of ADDS are discussed.

The effect of autism phenotype and diagnosis disclosure on students' desire for social distance from autistic peers.

Autistic students experience loneliness, rejection from peers, which might negatively affect both their well-being, as well as academic results. Others have studied this topic, however, the existing research does not analyze the desire for social distance from autistic female students in higher education. The goal of this study was to evaluate whether the way autism is expressed and disclosing the autism diagnosis had an effect on students' willingness to interact with autistic male and female students. We also analyzed participants' knowledge about autism, contact with autistic people in the past and its pleasantness in relation to their willingness to interact with autistic students described in the scenarios of the present study. We found that students were less willing to interact with autistic male students compared to autistic female students, as well as with autistic students whose diagnosis was not disclosed compared to autistic students whose diagnosis was disclosed to the raters. In addition, students who reported more pleasant contact with autistic individuals in the past were more willing to socially interact with autistic students if their diagnosis was disclosed. Students with higher knowledge of autism expressed greater willingness to interact with autistic males and autistic students regardless of diagnosis disclosure. Findings suggest that autistic males and individuals who prefer not to disclose their diagnosis are more vulnerable to social exclusion. More subtle expressions of autism need to be researched further.

Multi-Technique Characterization of 3.45 Ga Microfossils on Earth: A Key Approach to Detect Possible Traces of Life in Returned Samples from Mars.

The NASA Mars 2020 Perseverance rover is actively exploring Jezero crater to conduct analyses on igneous and sedimentary rock targets from outcrops located on the crater floor (Máaz and Séítah formations) and from the delta deposits, respectively. The rock samples collected during this mission will be recovered during the Mars Sample Return mission, which plans to bring samples back to Earth in the 2030s to conduct in-depth studies using sophisticated laboratory instrumentation. Some of these samples may contain traces of ancient martian life that may be particularly difficult to detect and characterize because of their morphological simplicity and subtle biogeochemical expressions. Using the volcanic sediments of the 3.45 Ga Kitty's Gap Chert (Pilbara, Australia), containing putative early life forms (chemolithotrophs) and considered as astrobiological analogues for potential early Mars organisms, we document the steps required to demonstrate the syngenicity and biogenicity of such biosignatures using multiple complementary analytical techniques to provide information at different scales of observation. These include sedimentological, petrological, mineralogical, and geochemical analyses to demonstrate macro- to microscale habitability. New approaches, some unavailable at the time of the original description of these features, are used to verify the syngenicity and biogenicity of the purported fossil chemolithotrophs. The combination of elemental (proton-induced X-ray emission spectrometry) and molecular (deep-ultraviolet and Fourier transform infrared) analyses of rock slabs, thin sections, and focused ion beam sections reveals that the carbonaceous matter present in the samples is enriched in trace metals (e.g., V, Cr, Fe, Co) and is associated with aromatic and aliphatic molecules, which strongly support its biological origin. Transmission electron microscopy observations of the carbonaceous matter documented an amorphous nanostructure interpreted to correspond to the degraded remains of microorganisms and their by-products (extracellular polymeric substances, filaments…). Nevertheless, a small fraction of carbonaceous particles has signatures that are more metamorphosed. They probably represent either reworked detrital biological or abiotic fragments of mantle origin. This study serves as an example of the analytical protocol that would be needed to optimize the detection of fossil traces of life in martian rocks.

Bodily-tactile early intervention: a pilot study of the role of maternal touch and emotional availability in interactions between three children with visual impairment and additional disabilities and their mothers.

Children with visual impairment and additional disabilities (VIAD) have difficulty accessing the visual information related to their parents' facial expressions and gestures. Similarly, it may be hard for parents to detect their children's subtle expressions. These challenges in accessibility may compromise emotional availability (EA) in parent-child interactions. The systematic use of the bodily-tactile modality for expressive and receptive communicative functions may function as a strategy to compensate for a child's lack of vision. This multiple-case study explored the effects of a bodily-tactile early intervention for three mothers and their one-year-old children with VIAD. Video data from baseline, intervention, and follow-up sessions were analyzed using a bodily-tactile coding procedure and EA Scales. During the intervention, all mothers began to use a more bodily-tactile modality in early play routines and in different communicative functions. They increased their use of anticipatory cues, noticing responses, and tactile signs. Moreover, the children were more emotionally available to their mothers during the intervention and follow-up compared to the baseline. The results indicated that, during a short intervention, mothers could adopt a systematic use of the bodily-tactile modality in interactions with their children with VIAD. The results also suggest that, when mothers increased flexibility in communication channels, it was positively linked to their children's EA.

Potential cyberbullying detection in social media platforms based on a multi-task learning framework

The proliferation of online violence has given rise to a spate of malignant incidents, necessitating a renewed focus on the identification of cyberbullying comments. Text classification lies at the heart of efforts to tackle this pernicious problem. The identification of cyberbullying comments presents unique challenges that call for innovative solutions. In contrast to traditional text classification tasks, cyberbullying comments are often accompanied by subtle and arbitrary expressions that can confound even the most sophisticated classification networks, resulting in low recognition accuracy and effectiveness. To address this challenge, a novel approach is proposed that leverages the BERT pre-training model for word embedding to retain the hidden semantic information in the text. Building on this foundation, the BiSRU++ model which combines attentional mechanisms is used to further extract contextual features of comments. A multi-task learning framework is employed for joint training of sentiment analysis and cyberbullying detection to improve the model's classification accuracy and generalization ability. The proposed model is no longer entirely reliant on a sensitive word dictionary, and experimental results demonstrate its ability to better understand semantic information compared to traditional models, facilitating the identification of potential online cyberbullying comments.