Pattern Discrimination Research Articles

Development of language-specific stress discrimination in European Portuguese: an electrophysiological study.

European Portuguese (EP) is a language with unpredictable stress. Previous behavioral studies have shown that without vowel reduction EP adult speakers displayed a stress deafness effect akin to that observed in speakers of fixed-stress languages, suggesting that vowel quality may be the primary cue for stress discrimination in EP. However, an event-related potentials (ERPs) study reported that EP adults were able to discriminate stress contrasts pre-attentively in the absence of vowel quality cues. These results seemed to indicate that EP adult speakers may attend to different cues in the attentive and pre-attentive stress perception. Moreover, both the behavioral and ERPs studies have revealed a processing advantage for iambic stress, which could not be predicted by the rhythmic properties of EP, the language-specific weighting of stress correlates, or the frequency distributions of trochaic and iambic stresses in EP. A recent eye-tracking study has found that EP-learning infants at 5-6 months already exhibited an iambic preference in the absence of vowel reduction, manifested by longer looking time at the iambic stress. The present study used a passive oddball paradigm to examine pre-attentive stress perception without vowel quality cues by 5-to-7-month-old EP-learning infants. Results from twenty-two participants showed that both the trochaic and iambic conditions yielded a positive discrimination response (p-MMR). In addition, the iambic condition elicited a prominent late discriminative negativity (LDN) as well as a P3a component. Our findings present the first evidence for reciprocal discrimination of stress patterns in EP-learning infants, showing that, as in adult speakers, stress processing might also differ at the pre-attentive and attentive stages in infants. Importantly, the stress perception ability in EP-learning infants seems to develop asymmetrically, with an advantage for the iambic stress pattern. The present study highlighted the role of language-specific factors that may affect developing stress perception.

Discriminative feature analysis of dairy products based on machine learning algorithms and Raman spectroscopy

Discriminant analysis of similar food samples is an important aspect of achieving food quality control. The effective combination of Raman spectroscopy and machine learning algorithms has become an extremely attractive approach to develop intelligent discrimination techniques. Feature spectral analysis can help researchers gain a deeper understanding of the data patterns in food quality discrimination. Herein, this work takes the discrimination of three brands of dairy products as an example to investigate the Raman spectral feature based on the support vector machines (SVM), extreme learning machines (ELM) and convolutional neural network (CNN) algorithms. The results show that there are certain differences in the optimal spectral feature interval corresponding to different machine learning algorithms. Selecting the appropriate spectral feature interval can maintain high recognition accuracy and improve the computational efficiency of the algorithm. For example, the SVM algorithm has a recognition accuracy of 100% in the 890-980 cm−1, 1410-1500 cm−1 fusion spectral range, which takes about 200 s. The ELM algorithm also has a recognition accuracy of 100% in the 890-980 cm−1, 1410-1500 cm−1 fusion spectral range, which takes less than 0.3 s. The CNN algorithm has a recognition accuracy of 100% in the 890-980 cm−1, 1050-1180 cm−1, 1410-1500 cm−1 fusion spectral range, which takes about 80 s. In addition, by analyzing the distribution of spectral feature intervals based on Euclidean distance, the distribution of experimental samples based on feature spectra is visually displayed. Through the spectral feature analysis process of similar samples, a set of analysis strategies is provided to deeply reveal the data foundation of classification algorithms, which can provide reference for the analysis of relevant discriminative research patterns.

CTD-Global (CTD-G): A novel composition, transition, and distribution based peptide sequence encoder for hormone peptide prediction

Hormone peptides are small signaling molecules that regulate key cellular processes such as cell growth, and differentiation. Hormone peptide identification is important for understanding their potential associations with certain diseases such as attention deficit hyperactivity disorder, diabetes, and psychiatric disorders. A comprehensive understanding of hormone peptides’ roles in cellular signaling, and immune regulation can provide insights into their therapeutic potential. Hormone peptides are identified through wet-lab approaches which are restricted by resource-intensive processes, limited scalability, and cost ineffectiveness. In an effort to substitute experimental approaches with computational predictors, researchers leveraged the capabilities of machine learning (ML) classifiers. These classifiers have inherent dependency over statistical vectors that are generated by extracting amino acids’ distinctive patterns from peptide sequences. Classifiers utilize these vectors for discriminating peptides into hormone and non-hormone classes. However, the performance of current predictors is constrained due to their inability to effectively extract discriminative amino acids patterns from peptide sequences. Following the need for a powerful predictor, the paper in hand presents a novel sequence encoder namely, CTD-G that transforms peptide sequences into statistical vectors by extracting 3 different types of amino acids patterns namely composition, transition, and distribution. Across public benchmark dataset, the proposed CTD-G encoder potential is compared with 56 existing encoders under two different evaluation strategies namely intrinsic and extrinsic. In Intrinsic evaluation, TSNE-based visualization demonstrates reduced overlap between clusters of hormone and non-hormone peptides with the proposed encoder’s statistical vectors compared to existing encoders. Extrinsic evaluation demonstrates the superiority of the proposed encoder, as 7 out of 11 ML classifiers achieve better performance with its statistical vectors compared to those from existing encoders. Furthermore, the proposed predictor outperforms existing hormone peptide classification predictors by 1.5% in accuracy, 5.36% in sensitivity, 1.80% in specificity, and 2.62% in MCC. To facilitate the scientific community, a web application is available at https://sds_genetic_analysis.opendfki.de/.

MECOM: A Meta-Completion Network for Fine-Grained Recognition With Incomplete Multi-Modalities.

Our work focuses on tackling the problem of fine-grained recognition with incomplete multi-modal data, which is overlooked by previous work in the literature. It is desirable to not only capture fine-grained patterns of objects but also alleviate the challenges of missing modalities for such a practical problem. In this paper, we propose to leverage a meta-learning strategy to learn model abilities of both fast modal adaptation and more importantly missing modality completion across a variety of incomplete multi-modality learning tasks. Based on that, we develop a meta-completion method, termed as MECOM, to perform multimodal fusion and explicit missing modality completion by our proposals of cross-modal attention and decoupling reconstruction. To further improve fine-grained recognition accuracy, an additional partial stream (as a counterpart of the main stream of MECOM, i.e., holistic) and the part-level features (corresponding to fine-grained objects' parts) selection are designed, which are tailored for fine-grained nature to capture discriminative but subtle part-level patterns. Comprehensive experiments from quantitative and qualitative aspects, as well as various ablation studies, on two fine-grained multimodal datasets and one generic multimodal dataset show our superiority over competing methods. Our code is open-source and available at https://github.com/SEU-VIPGroup/MECOM.

Discriminación y limitaciones en el uso de la lengua española: el fetiche de la norma en Twitter (X)

This paper discusses the restrictions that the standard norm of Spanish imposes on speakers. T he analysis has been based on opinions expressed on the social network Twitter regarding memes that present spelling or grammatical deviations from the standard norm of Spanish. Firstly, some terms are discussed to contextualize this phenomenon, such as linguistic profiling and linguistic discrimination (linguicism, accentism, and glottophobia). After that, a corpus of one hundred comments collected from Twitter between November and December 2022 is analyzed. This corpus serves as textuality to illustrate how society reproduces discriminatory patterns based on the assumption of regulations understood in terms of command and prestige. The main conclusion of this research is that the restrictive nature of the languages governed by normative institutions limits the innovative practices of a linguistic community and encourages discriminatory and classist practices.

Cancer evaluation in dogs using cerumen as a source for volatile biomarker prospection.

Cancer is one of the deadliest diseases in humans and dogs. Nevertheless, most tumor types spread faster in canines, and early cancer detection methods are necessary to enhance animal survival. Here, cerumen (earwax) was tested as a source of potential biomarkers for cancer evaluation in dogs. Earwax samples from dogs were collected from tumor-bearing and clinically healthy dogs, followed by Headspace/Gas Chromatography-Mass Spectrometry (HS/GC-MS) analyses and multivariate statistical workflow. An evolutionary-based multivariate algorithm selected 18 out of 128 volatile metabolites as a potential cancer biomarker panel in dogs. The candidate biomarkers showed a full discrimination pattern between tumor-bearing dogs and cancer-free canines with high accuracy in the test dataset: an accuracy of 95.0% (75.1-99.9), and sensitivity and specificity of 100.0% and 92.9%, respectively. In summary, this work raises a new perspective on cancer diagnosis in dogs, being carried out painlessly and non-invasive, facilitating sample collection and periodic application in a veterinary routine.

Boosting lower-limb motor imagery performance through an ensemble method for gait rehabilitation

Lower-limb exoskeletons have been used extensively in many rehabilitation applications to assist disabled people with their therapies. Brain–machine interfaces (BMIs) further provide effective and natural control schemes. However, the limited performance of brain signal decoding from lower-limb kinematics restricts the broad growth of both BMI and rehabilitation industry. To address these challenges, we propose an ensemble method for lower-limb motor imagery (MI) classification. The proposed model employs multiple techniques to boost performance, including deep and shallow parts. Traditional wavelet transformation followed by filter-bank common spatial pattern (CSP) employs neurophysiologically reasonable patterns, while multi-head self-attention (MSA) followed by temporal convolutional network (TCN) extracts deeper encoded generalized patterns. Experimental results in a customized lower-limb exoskeleton on 8 subjects in 3 consecutive sessions showed that the proposed method achieved 60.27% and 64.20% for three (MI of left leg, MI of right leg, and rest) and two classes (lower-limb MI vs. rest), respectively. Besides, the proposed model achieves improvements of up to 4% and 2% accuracy for the subject-specific and subject-independent modes compared to the current state-of-the-art (SOTA) techniques, respectively. Finally, feature analysis was conducted to show discriminative brain patterns in each MI task and sessions with different feedback modalities. The proposed models integrated in the brain-actuated lower-limb exoskeleton established a potential BMI for gait training and neuroprosthesis.

Understand service animal requirements on campus

The Office for Civil Rights has published several resolution agreements/letters of findings in 2023 involving discrimination against individuals with disabilities who use service animals. These letters can help postsecondary institutions identify common patterns of discrimination in service animal policies and help them avoid compliance issues in their own policies. This article will summarize the major 2023 OCR letters that address service animals, Section 504, and Title II.

Study on the mechanical properties and acoustic emission signal characteristics of freezing pipe

The increase in freezing depth requires thicker, stronger, and colder freezing walls, and various complex factors in deep strata greatly increase the risk of freezing pipe fracture. To address the phenomenon of freezing pipe fracture, this paper designs freezing pipe and joint mechanical performance experiments based on acoustic emission (AE) technology, mainly testing the deformation of freezing steel pipes and composite joints at normal and low temperatures, changes in load bearing capacity, and corresponding AE characteristics of the process. Additionally, the associated AE characteristics throughout the process will be analyzed. The ultimate goal is to establish a discriminative pattern for identifying the critical fracture of freezing pipes based on the analysis of AE signal characteristics in conjunction with mechanical properties. The sensitivity of the AE system under low-temperature conditions and the reliability of the test were tested through pencil lead break experiments, and saltwater noise detection experiments were conducted to prevent noise interference from low-temperature saltwater flow and pipe wall friction in the identification of crucial signals for freezing pipe fracture. This study provides a basis for identifying the deformation mechanics and fracture warning of freezing pipes through dynamic analysis of AE monitoring information.

A hybrid spatiotemporal deep belief network and sparse representation-based framework reveals multilevel core functional components in decoding multitask fMRI signals.

Decoding human brain activity on various task-based functional brain imaging data is of great significance for uncovering the functioning mechanism of the human mind. Currently, most feature extraction model-based methods for brain state decoding are shallow machine learning models, which may struggle to capture complex and precise spatiotemporal patterns of brain activity from the highly noisy fMRI raw data. Moreover, although decoding models based on deep learning methods benefit from their multilayer structure that could extract spatiotemporal features at multiscale, the relatively large populations of fMRI datasets are indispensable, and the explainability of their results is elusive. To address the above problems, we proposed a computational framework based on hybrid spatiotemporal deep belief network and sparse representations to differentiate multitask fMRI (tfMRI) signals. Using a relatively small cohort of tfMRI data as a test bed, our framework can achieve an average classification accuracy of 97.86% and define the multilevel temporal and spatial patterns of multiple cognitive tasks. Intriguingly, our model can characterize the key components for differentiating the multitask fMRI signals. Overall, the proposed framework can identify the interpretable and discriminative fMRI composition patterns at multiple scales, offering an effective methodology for basic neuroscience and clinical research with relatively small cohorts.

RACIAL DISCRIMINATION TRAJECTORIES AND MULTIMORBIDITY BURDEN AND CHANGE AMONG MIDDLE-AGED AND OLDER BLACKS

Abstract Racial discrimination is a chronic psychosocial stressor that can accelerate aging. Characterizing longitudinal patterns of racial discrimination may provide greater precision in predicting health risk particularly for outcomes such as multimorbidity that is considered a proxy of multisystem dysregulation and a marker of cumulative disease burden. However, longitudinal patterns of racial discrimination and its relationship with multimorbidity has received limited attention. We investigate the association between racial discrimination trajectories (RDT) and multimorbidity among a subsample of middle-aged and older Black adults from the Health and Retirement Study (2006–2020, N = 2,167, 50+). RDTs were constructed based on the Everyday Discrimination Scale using repeated measures latent profile analyses. Multimorbidity was operationalized using the multimorbidity weighted index (MWI) score. Linear mixed models, adjusted for sociodemographic, behavioral, and clinical covariates, estimated the association between RDT and MWI score and change in score over time. Average MWI scores varied across RDT classes: low (5.9), moderate (6.5), and high (5.9). Individuals characterized in the moderate RDT had the highest multimorbidity burden (3.03; 95% confidence interval [CI]: 1.46, 4.58) and was the only group that experienced an accelerated rate of multimorbidity accumulation, as measured by change in MWI score over time. This relationship was not observed among respondents in the persistently high RDT. Our findings suggest that there are differences in the relationship between RDT and multimorbidity burden. Uncovering specific characteristics associated with the profiles of RDT may advance our understanding of healthy aging processes.

AGE DIFFERENCES IN EVERYDAY DISCRIMINATION AND CORTISOL DYNAMIC RANGE IN DAILY LIFE

Abstract Extensive work has linked everyday discrimination with poorer health. However, research testing age differences in this relationship is limited. The current study examined everyday discrimination and cortisol dynamic range, and whether the association differed by age. Cortisol dynamic range is believed to reflect physiological responsiveness to external demands such as stressors, with a compressed range indicating less responsiveness. Participants were 250 community-based adults living in British Columbia (ages 25-88, mean=46 years, 64% White, 68% women). At baseline, participants completed the Everyday Discrimination Scale, followed by four days of at-home saliva collection 3x per day: upon waking, 30 minutes post-waking, and before bed. Cortisol dynamic range was calculated as the log-cortisol peak minus log-cortisol nadir across the four days. Multiple regression controlled for race/ethnicity, education, gender, medications, average length of waking day, and depressive symptoms. Results showed a negative association between age and cortisol dynamic range, such that older age was associated with a smaller range (b=-0.01, SE=0.003, p<0.001). There was no main effect of everyday discrimination on cortisol dynamic range. However, there was a significant interaction between age and everyday discrimination. Among younger adults, more frequent everyday discrimination was associated with a smaller cortisol dynamic range. In contrast, among middle-aged adults, more frequent everyday discrimination was associated with a larger range (b=0.37, SE=0.08, p<0.001). There was no significant association between discrimination and cortisol dynamic range among older adults. Findings indicate differing patterns in discrimination and cortisol dynamic range between younger, middle-aged, and older adults.

Queers and Right-wing Ideology: Case Study of Serbia

The systematic exclusion of trans people from society vividly demonstrates the oppression faced by marginalised groups. These patterns of discrimination are also seen in instances of hatred towards women or non-heterosexual individuals. Within the queer community, trans people are particularly vulnerable and are often the target of conservative and right-wing movements. It is crucial to acknowledge that the oppression of one group can have a domino effect on the entire LGBTIQ+ community. It is noteworthy that patterns of oppression cannot be observed in isolation, since neoliberal policies tend to fragment solidarity and our common struggles. Despite not being immediately apparent, the entire queer community suffers the consequences of the exclusion of trans people. Similarly, policies of segregation and exclusion based on sexual orientation and gender identity have broader consequences, which may also impact the rights and freedoms of heterosexual cisgender individuals. Right-wing policies and ideology tend to spread, and this can be seen in the example of Serbia, where institutional and non-institutional levels witness increased activity of right-wing groups.

Invisibility: From Discrimination to Resistance

ABSTRACTThe paper takes heed of the fact that, when evaluating normative issues through the semantics of visibility and invisibility, a transfer takes place from optical to political semantics which is not without consequences. The paper attempts a typology of the extremely diverse functions (in)visibility takes in current discourses, moving from the characterization of situations of discrimination to that of the resistance to it. In a first step, it analyses the affirmative uses of the notion of political visibilisation, whether of individuals, groups or concerns, that fashion themselves as targeted answers to discrimination. In a second step, it looks at the politics of invisibilisation, stealth, and opaqueness, claiming emancipatory and critical potential in resisting the limelight of public space. Specifically, it assesses whether the traditional connect of politicization and public visibility still holds true in all cases. In a third part, the paper suggests a possible reorganization of the polarity of visibility and invisibility as one that runs along the continuum between determinacy and indeterminacy, indicating its effects for patterns of discrimination as well as for practices of resistance.

Is algorithmic credit scoring a ‘high risk’?

Current proposals of European regulations — the EU Artificial Intelligence Act and the UK pro-innovation approach to regulating artificial intelligence (AI) — either define credit scoring as a ‘high-risk’ application or mention credit scoring in the context of issues of ‘fairness’. Therefore, the question of whether algorithmic credit scoring is a ‘high risk’ might be a rhetorical one. Banks and lenders, however, have to comply with these proposed (and probably upcoming) regulations and explain that the current (and future) algorithms are compliant. This paper will use the — existing and indisputable — ‘gender pay gap’ together with a toy model of credit scoring based on income alone to discuss the regulatory requirements and the potential statistical methods to prove compliance. Although these regulations are triggered by the developments of ‘artificial intelligence’, all such tolls are nothing more than ‘statistical classifier’ and can be discussed — in a nutshell and with many simplifications — using this toy model. A closer examination of the statistical concepts reveals that simple correlations alone (eg a test of whether any outcome is independent of a sensitive parameter ‘gender’) cannot provide any answer about any ‘perpetuation of historical patterns of discrimination’ or any ‘fairness’. It requires advanced statistical methods (based on the concepts of mitigation and causality) to show that credit scoring is fully gender agnostic and, consequently, poses no risk to fundamental rights, as long as similar economic situations generally result in similar outcomes (score value or credit approvals). This argument, however, requires an understanding of advanced statistical approaches, and common misunderstandings can expose lenders to reputational risk. While banks and lenders have to use such advanced approaches to prove compliance, they have, in parallel, to communicate the underlying causality to the stakeholders in an understandable manner.

Visual rating scales of atrophy to differentiate Frontotemporal dementia from Primary psychiatric disorder: results from DIPPA study

AbstractBackgroundAdult‐onset behavioral changes and altered executive functioning are frequently caused by behavioural variant of Frontotemporal dementia (bvFTD) or primary psychiatric disorders (PPD) which overlap in terms of clinical presentations but differ in terms of treatment and prognosis. The multi‐centre study DIPPA‐FTD aims to develop diagnostic and prognostic algorithms to help distinguish sporadic bvFTD from late‐onset PPD. Although atrophy on neuroimaging seems to be an appropriate discriminator between neurodegenerative and non‐neurodegenerative conditions, the sensitivity of frontotemporal atrophy for bvFTD is relatively low, whereas decreased brain volumes may be found in conditions like schizophrenia. The aim of the study is to identify a discriminative pattern of brain atrophy between bvFTD and PPD in a clinically applicable way.MethodThe DIPPA FTD consortium retrospectively collected patients with late onset behavioral disturbances after the age of 45 years, across 5 centers (Milan, Amsterdam, Munich, Sydney, Montreal). The subjects had been classified either as bvFTD or PPD according to current clinical criteria and the majority had clinical follow‐up. Among the patients collected in the project, only those with T1 MRI available were selected. A protocol of 9 visual rating scales of atrophy (orbitofrontal, anterior cingulate, anterior temporal, fronto‐insula, medial temporal, parietal, medial ventricular and axial ventricular) was applied by a rater blind for clinical and demographic infomation. The rater was asked to classify the subjects as bvFTD or PPD. A composite score of frontal, temporal and ventricular rating scales was also calculated.ResultThe MRIs of 323 subjects (211 bvFTD and 112 PPD) were analysed. Groupwise bvFTD cases showed significantly higher scores of atrophy for all the scales used. The rater accurately predicted 72% of the cases (Sens 0.70, Spec 0.76) while the composite score reached an accuracy of 76% (Sens 0.77 Spec 0.74). ROC curve analysis showed that fronto‐insula was the single most useful scale in the differentiation between bvFTD and PPD (AUC 0.80).ConclusionBrain atrophy has a significant role in the discrimination between bvFTD and PPD. The use of visual rating scales, and in particular the fronto‐insula one, could increase the diagnostic accuracy in the clinical setting.

Visual rating scales of atrophy to differentiate Frontotemporal dementia from Primary psychiatric disorder: results from DIPPA study

AbstractBackgroundAdult‐onset behavioral changes and altered executive functioning are frequently caused by behavioural variant of Frontotemporal dementia (bvFTD) or primary psychiatric disorders (PPD) which overlap in terms of clinical presentations but differ in terms of treatment and prognosis. The multi‐centre study DIPPA‐FTD aims to develop diagnostic and prognostic algorithms to help distinguish sporadic bvFTD from late‐onset PPD. Although atrophy on neuroimaging seems to be an appropriate discriminator between neurodegenerative and non‐neurodegenerative conditions, the sensitivity of frontotemporal atrophy for bvFTD is relatively low, whereas decreased brain volumes may be found in conditions like schizophrenia. The aim of the study is to identify a discriminative pattern of brain atrophy between bvFTD and PPD in a clinically applicable way.MethodThe DIPPA FTD consortium retrospectively collected patients with late onset behavioral disturbances after the age of 45 years, across 5 centers (Milan, Amsterdam, Munich, Sydney, Montreal). The subjects had been classified either as bvFTD or PPD according to current clinical criteria and the majority had clinical follow‐up. Among the patients collected in the project, only those with T1 MRI available were selected. A protocol of 9 visual rating scales of atrophy (orbitofrontal, anterior cingulate, anterior temporal, fronto‐insula, medial temporal, parietal, medial ventricular and axial ventricular) was applied by a rater blind for clinical and demographic infomation. The rater was asked to classify the subjects as bvFTD or PPD. A composite score of frontal, temporal and ventricular rating scales was also calculated.ResultThe MRIs of 323 subjects (211 bvFTD and 112 PPD) were analysed. Groupwise bvFTD cases showed significantly higher scores of atrophy for all the scales used. The rater accurately predicted 72% of the cases (Sens 0.70, Spec 0.76) while the composite score reached an accuracy of 76% (Sens 0.77 Spec 0.74). ROC curve analysis showed that fronto‐insula was the single most useful scale in the differentiation between bvFTD and PPD (AUC 0.80).ConclusionBrain atrophy has a significant role in the discrimination between bvFTD and PPD. The use of visual rating scales, and in particular the fronto‐insula one, could increase the diagnostic accuracy in the clinical setting.

A hybrid graph network model for ASD diagnosis based on resting-state EEG signals

Autism spectrum disorder (ASD) is a common neurodevelopmental disorder and early diagnosis is crucial for effective treatment. Stable and effective biomarkers are essential for understanding the underlying causes of the disorder and improving diagnostic accuracy. Electroencephalography (EEG) signals have proven to be reliable biomarkers for diagnosing ASD. Extracting stable connectivity patterns from EEG signals helps ensure robustness in ASD diagnostic systems. In this study, we propose a hybrid graph convolutional network framework called Rest-HGCN, which utilizes resting-state EEG signals to capture differential patterns of brain connectivity between normal children and ASD patients using graph learning strategies. The Rest-HGCN combines brain network analysis techniques and data-driven strategies to extract discriminative graph features from resting-state EEG signals. By automatically extracting differential graph patterns from these signals, the Rest-HGCN achieves reliable ASD diagnosis. To evaluate the performance of Rest-HGCN, we conducted ASD diagnosis experiments using k-fold cross-validation on the public ABC-CT resting EEG dataset. The proposed Rest-HGCN model achieved accuracies of 87.12 % and 85.32 % in single-subject and cross-experiment analyses, respectively. The results suggest that Rest-HGCN can effectively capture discriminant graph patterns from resting EEG signals and achieve robust ASD diagnosis. This may provide an effective and convenient tool for clinical ASD diagnosis.

Passion-Net: a robust precise and explainable predictor for hate speech detection in Roman Urdu text

With an aim to eliminate or reduce the spread of hate content across social media platforms, the development of artificial intelligence supported computational predictors is an active area of research. However, diversity of languages hinders development of generic predictors that can precisely identify hate content. Several language-specific hate speech detection predictors have been developed for most common languages including English, Chinese and German. Specifically, for Urdu language a few predictors have been developed and these predictors lack in predictive performance. The paper in hand presents a precise and explainable deep learning predictor which makes use of advanced language modelling strategies for the extraction of semantic and discriminative patterns. Extracted patterns are utilized to train an attention-based novel classifier that is competent in precisely identifying hate content. Over coarse-grained benchmark dataset, the proposed predictor significantly outperforms state-of-the-art predictor by 8.7% in terms of accuracy, precision and F1-score. Similarly, over fine-grained dataset, in comparison with state-of-the-art predictor, it achieves performance gain of 10.6%, 17.6%, 18.6% and 17.6% in terms of accuracy, precision, recall and F1-score.

Developing System-Based Artificial Intelligence Models for Detecting the Attention Deficit Hyperactivity Disorder

This study presents a novel methodology for automating the classification of pediatric ADHD using electroencephalogram (EEG) biomarkers through machine learning and deep learning techniques. The primary objective is to develop accurate EEG-based screening tools to aid clinical diagnosis and enable early intervention for ADHD. The proposed system utilizes a publicly available dataset consisting of raw EEG recordings from 61 individuals with ADHD and 60 control subjects during a visual attention task. The methodology involves meticulous preprocessing of raw EEG recordings to isolate brain signals and extract informative features, including time, frequency, and entropy signal characteristics. The feature selection techniques, including least absolute shrinkage and selection operator (LASSO) regularization and recursive elimination, were applied to identify relevant variables and enhance generalization. The obtained features are processed by employing various machine learning and deep learning algorithms, namely CatBoost, Random Forest Decision Trees, Convolutional Neural Networks (CNNs), and Long Short-Term Memory Networks (LSTMs). The empirical results of the proposed algorithms highlight the effectiveness of feature selection approaches in matching informative biomarkers with optimal model classes. The convolutional neural network model achieves superior testing accuracy of 97.75% using LASSO-regularized biomarkers, underscoring the strengths of deep learning and customized feature optimization. The proposed framework advances EEG analysis to uncover discriminative patterns, significantly contributing to the field of ADHD screening and diagnosis. The suggested methodology achieved high performance compared with different existing systems based on AI approaches for diagnosing ADHD.