Latent Domain Research Articles

Adverse events related to antiepileptic drugs

ObjectiveAdverse events (AEs) related to antiepileptic drugs (AEDs) may interfere with adequate dosing and patient adherence, leading to suboptimal seizure control, and relatedly, increased injuries, hospitalizations, and mortality. This study investigated the clinicodemographic factors associated with AEs related to AEDs as reported by the Liverpool Adverse Events Profile (LAEP), and explored the ability of LAEP to discriminate between epilepsy and psychogenic nonepileptic seizures (PNES). We hypothesized that female sex, mood disorders, AED-polytherapy, duration, and severity of epilepsy are associated with increased endorsement of AEs related to AEDs, and that endorsement of AEs related to AEDs would significantly differ between epilepsy and PNES patients. MethodsWe prospectively enrolled adult patients admitted to two inpatient video-electroencephalogram monitoring units. Clinicodemographic variables and psychometric measures of depression, anxiety, and cognitive function were recorded. Patient-reported AE endorsement was obtained using the LAEP, which was reduced to four latent domains using exploratory structural equation modeling. General linear models identified variables associated with each domain. Logistic regression determined the ability of LAEP scores to differentiate between epilepsy and PNES. Results311 patients met inclusion criteria. Mean age was 38 years and 56% of patients were female. Network analysis demonstrated strong relationships between depression and anxiety with physical, sleep, psychiatric, and dermatological AE endorsement. Depression, female sex, and AED polytherapy were associated with greater AE endorsement. Epilepsy, compared to PNES, was associated with lower AE endorsement. Fewer prescribed AEDs and greater reported physical AE endorsement were associated with PNES diagnosis. SignificanceThere is a strong relationship between patient-reported AEs and psychiatric symptomatology. Those with PNES paradoxically endorse greater physical AEs despite receiving fewer AEDs. Patients who endorse AEs in clinical practice should be screened for comorbid depression or anxiety and treated accordingly.

Transfer Learning for Wireless Fingerprinting Localization Based on Optimal Transport.

Wireless fingerprinting localization (FL) systems identify locations by building radio fingerprint maps, aiming to provide satisfactory location solutions for the complex environment. However, the radio map is easy to change, and the cost of building a new one is high. One research focus is to transfer knowledge from the old radio maps to a new one. Feature-based transfer learning methods help by mapping the source fingerprint and the target fingerprint to a common hidden domain, then minimize the maximum mean difference (MMD) distance between the empirical distributions in the latent domain. In this paper, the optimal transport (OT)-based transfer learning is adopted to directly map the fingerprint from the source domain to the target domain by minimizing the Wasserstein distance so that the data distribution of the two domains can be better matched and the positioning performance in the target domain is improved. Two channel-models are used to simulate the transfer scenarios, and the public measured data test further verifies that the transfer learning based on OT has better accuracy and performance when the radio map changes in FL, indicating the importance of the method in this field.

Multi-TALK: Multi-Microphone Cross-Tower Network for Jointly Suppressing Acoustic Echo and Background Noise.

In this paper, we propose a multi-channel cross-tower with attention mechanisms in latent domain network (Multi-TALK) that suppresses both the acoustic echo and background noise. The proposed approach consists of the cross-tower network, a parallel encoder with an auxiliary encoder, and a decoder. For the multi-channel processing, a parallel encoder is used to extract latent features of each microphone, and the latent features including the spatial information are compressed by a 1D convolution operation. In addition, the latent features of the far-end are extracted by the auxiliary encoder, and they are effectively provided to the cross-tower network by using the attention mechanism. The cross tower network iteratively estimates the latent features of acoustic echo and background noise in each tower. To improve the performance at each iteration, the outputs of each tower are transmitted as the input for the next iteration of the neighboring tower. Before passing through the decoder, to estimate the near-end speech, attention mechanisms are further applied to remove the estimated acoustic echo and background noise from the compressed mixture to prevent speech distortion by over-suppression. Compared to the conventional algorithms, the proposed algorithm effectively suppresses the acoustic echo and background noise and significantly lowers the speech distortion.

A convergent framework with learnable feasibility for Hadamard-based image recovery

In this paper, we propose a framework for recovering image degradations that can be formulated by the Hadamard product of clear images with degradation factors. By training the mapping from datasets, we show that implicit feasibilities can be learned in forms of latent domains. Then with the feasibilities and acknowledged data priors, the recovery problems are formulated as a general optimization model in which the domain knowledge of degradations are also nicely involved. Then we solve the model based on the classical coordinate update with plugged-in networks so that all the variables can be well estimated. Even better, our updating scheme is designed under the guidance of theoretical analyses, thus its stability can always be guaranteed in practice. We show that different recovery problems can be solved under our unified framework, and the extensive experimental results verify that the proposed framework is superior to state-of-the-art methods in both benchmark datasets and real-world images.

Examining determinants of gender attitudes: evidence among Tanzanian adolescents

BackgroundThe shaping of gender beliefs and attitudes in early adolescence affects the way young people internalize and self-enforce prevalent notions of masculinity and femininity, with lifelong consequences for sexual and reproductive health. This cross-sectional study examines determinants of gender attitudes among some of the poorest and most vulnerable adolescents in Tanzania using an ecological model.MethodsData come from baseline interviews with 2458 males and females aged 14–19 years conducted as part of a larger impact evaluation. Structural equation models are used to examine how factors at the community-, household-, and individual-levels influence gender attitudes in the four domains measured by the Gender Equitable Men (GEM) Scale (i.e. violence, sexual relationships, reproductive health and disease prevention, and domestic chores and daily life).ResultsA structural equation model of the four latent domains of the GEM scale regressed on individual, social-interactional and structural level characteristics indicated that secondary school attendance was associated with more equitable gender attitudes, while females held less equitable attitudes than males in the sample. Having had sexual intercourse was associated with more gender equitable attitudes among females, but the reverse was true among males.ConclusionsAddressing gender inequity requires understanding gender socialisation at the socio-interactional level. As females had more inequitable gender attitudes than males in the study, a special emphasis on highlighting the rights of women to girls should be considered. This study will inform future analysis of programme impacts on gender attitudes and sexual and reproductive health.

Is there a G factor for metacognition? Correlations in retrospective metacognitive sensitivity across tasks.

Is metacognition a general resource shared across domains? Previous research has documented consistent biases in judgments across tasks. In contrast, there is debate regarding the domain generality or the domain specificity of the ability to discriminate between correct and incorrect answers (metacognitive sensitivity) because most previous work has documented nonsignificant correlations across domains. However, such null findings may be due to low statistical power and differences in task structure or performance, thereby masking a latent domain generality in metacognition. We examined across-domain correlations in confidence level and sensitivity in a large sample (N = 181). Participants performed 4 2-alternative forced-choice tasks (episodic memory, semantic memory, executive function, and visual perception) with trial-by-trial confidence judgments. We found significant correlations in average confidence level across tasks. By applying a hierarchical Bayesian model to estimate cross-task covariance, we found five out 6 cross-task correlations in metacognitive efficiency (meta-d′/d′) were significant, even for pairs of tasks in which first-order performance was not correlated. This suggests that at least some components of metacognitive efficiency in retrospective confidence are domain general.

Discovering and incorporating latent target-domains for domain adaptation

In this paper, we aim to address the unsupervised domain adaptation problem where the data in the target domain are much more diverse compared with the data in the source domain. In particular, this problem is formulated as discovering and incorporating latent domains underlying target data of interest for unsupervised domain adaptation. More specifically, the discovery of the latent target domains is based on three criteria, including the maximization of compactness and distinctiveness of the data in the individual latent target-domain, as well as the minimization of total divergence from the latent target-domains to the source domain. For each pair formed by a latent target domain and the source domain, we learn a feature space where the discrepancy between the source domain and the specific latent target domain is shrunk. Finally, we consider the projected source domain data on the learned latent feature spaces as different views of the source domain, and propose an extended multiple kernel learning algorithm to train a more robust and precise classifier for predicting the unlabeled target data. The effectiveness of our proposed method is demonstrated on various benchmark datasets for object recognition and human activity recognition. Moreover, we also show that our proposed method can be treated as an effective complement to the deep learning based unsupervised domain adaptation.

Transfer Metric Learning for Unseen Domains

We propose a transfer metric learning method to infer domain-specific data embeddings for unseen domains, from which no data are given in the training phase, by using knowledge transferred from related domains. When training and test distributions are different, the standard metric learning cannot infer appropriate data embeddings. The proposed method can infer appropriate data embeddings for the unseen domains by using latent domain vectors, which are latent representations of domains and control the property of data embeddings for each domain. This latent domain vector is inferred by using a neural network that takes the set of feature vectors in the domain as an input. The neural network is trained without the unseen domains.The proposed method can instantly infer data embeddings for the unseen domains without (re)-training once the sets of feature vectors in the domains are given. To accumulate knowledge in advance, the proposed method uses labeled and unlabeled data in multiple source domains. Labeled data, i.e., data with label information such as class labels or pair (similar/dissimilar) constraints, are used for learning data embeddings in such a way that similar data points are close and dissimilar data points are separated in the embedding space. Although unlabeled data do not have labels, they have geometric information that characterizes domains. The proposed method incorporates this information in a natural way on the basis of a probabilistic framework.The conditional distributions of the latent domain vectors, the embedded data, and the observed data are parameterized by neural networks and are optimized by maximizing the variational lower bound using stochastic gradient descent. The effectiveness of the proposed method was demonstrated through experiments using three clustering tasks.

S237. MODERATORS AND RELATIONSHIPS OF COGNITION AND SUBJECTIVE QUALITY OF LIFE IN SCHIZOPHRENIA

BackgroundThe relationship between cognitive impairments and subjective quality of life (sQOL) in schizophrenia remains indeterminate, relative to studies of objective QOL (oQOL), despite much previous work. This study sought to better characterise the cognition-sQOL relationship in schizophrenia by 1) examining associations between factor analysis-derived cognitive domains and sQOL, 2) investigating if these domains predicted sQOL over other demographic and clinical variables, and 3) exploring if clinical, demographic and functional variables moderated the significant relationships.Methods47 schizophrenia/schizoaffective disorder patients (age: M=43.72, SD=11.35) and 48 healthy controls (age: M=39.82, SD=13.89) were assessed. QOL was measured using the Lehman’s QOL Interview. Composite scores were created to represent oQOL and sQOL, and factor analysis (principal axis factoring with varimax rotation) assessed for latent domains within 14 cognitive tasks. Symptomatology was assessed using the PANSS and MADRS. Pearson’s correlations and a stepwise linear regression were conducted to examine the associations between cognition and sQOL before the moderation analyses.ResultsThree cognitive domains were derived: visuospatial planning, verbal linguistic and inhibition switching. Only tasks comprising the verbal linguistic cognitive domain were significantly associated with, and predicted, sQOL (p<.05). Moderation analyses revealed that the direction of this relationship differed between patients and healthy controls (p<.05), and was moderated in patients only by positive symptom severity (p<.05).DiscussionThe relationship between cognition and sQOL in schizophrenia is determined by, and more associated with, meaning-based cognitions. It is affected by symptomatology, notably positive symptoms, which lead to worsening sQOL among those with more intact verbal linguistic processing.

Confirmatory Factor Analysis and Construct Validity of the Internalized Sex Work Stigma Scale among a Cohort of Cisgender Female Sex Workers in Baltimore, Maryland, United States

ABSTRACT Internalized sex work stigma among cisgender female sex workers (FSW) is produced within contexts of social marginalization and associated with a range of ill-effects, including psychological distress, and lower rates of healthcare-seeking. This study seeks to uncover latent domains of the new Internalized Sex Work Stigma Scale (ISWSS) using data from 367 FSW in Baltimore, Maryland, USA. The sample was 56% white with high substance use (82% smoked crack cocaine, 58% injected any drug). The average ISWSS score was 34.8 (s.d. = 5.8, possible range: 12–48) and internal consistency was high (0.82). Confirmatory factor analysis revealed four subscales: worthlessness, guilt and shame, stigma acceptance, and sex work illegitimacy. Internal consistency of subscales was high (0.69–0.90); the scale also demonstrated construct validity with depression and agency. In bivariate logistic regressions, higher ISWSS, worthlessness, shame and guilt, and acceptance scores predicted higher odds of rushing client negotiations due to police. In unadjusted multinomial regressions, feeling respected by police predicted lower ISWSS, worthlessness, guilt and shame, acceptance, and illegitimacy scores. Identified factors are congruent with existing literature about how FSW manage sex work-specific stigma. Understanding the unique dimensions and impacts of internalized sex work stigma can inform interventions and policy to reduce morbidities experienced by FSW.

Domain Generalization Using a Mixture of Multiple Latent Domains

When domains, which represent underlying data distributions, vary during training and testing processes, deep neural networks suffer a drop in their performance. Domain generalization allows improvements in the generalization performance for unseen target domains by using multiple source domains. Conventional methods assume that the domain to which each sample belongs is known in training. However, many datasets, such as those collected via web crawling, contain a mixture of multiple latent domains, in which the domain of each sample is unknown. This paper introduces domain generalization using a mixture of multiple latent domains as a novel and more realistic scenario, where we try to train a domain-generalized model without using domain labels. To address this scenario, we propose a method that iteratively divides samples into latent domains via clustering, and which trains the domain-invariant feature extractor shared among the divided latent domains via adversarial learning. We assume that the latent domain of images is reflected in their style, and thus, utilize style features for clustering. By using these features, our proposed method successfully discovers latent domains and achieves domain generalization even if the domain labels are not given. Experiments show that our proposed method can train a domain-generalized model without using domain labels. Moreover, it outperforms conventional domain generalization methods, including those that utilize domain labels.

Visual Domain Adaptation by Consensus-Based Transfer to Intermediate Domain

We describe an unsupervised domain adaptation framework for images by a transform to an abstract intermediate domain and ensemble classifiers seeking a consensus. The intermediate domain can be thought as a latent domain where both the source and target domains can be transferred easily. The proposed framework aligns both domains to the intermediate domain, which greatly improves the adaptation performance when the source and target domains are notably dissimilar. In addition, we propose an ensemble model trained by confusing multiple classifiers and letting them make a consensus alternately to enhance the adaptation performance for ambiguous samples. To estimate the hidden intermediate domain and the unknown labels of the target domain simultaneously, we develop a training algorithm using a double-structured architecture. We validate the proposed framework in hard adaptation scenarios with real-world datasets from simple synthetic domains to complex real-world domains. The proposed algorithm outperforms the previous state-of-the-art algorithms on various environments.

Detecting malware communities using socio-cultural cognitive mapping

We apply a variation of socio-cultural cognitive mapping (SCM) to computer malware features explored previously by Saxe and Berlin that characterized malware binaries as benign or malicious based on 1024 program features derived from a deep neural network-based detection system. In this work, we model the features as attributes within a latent spatial domain using a weighted consensus graph representation to visualize and analyze the malware binary communities. The data used in our analysis is extracted from a Remote Access Trojan family named Sakula that first appeared in 2012, and has been used to enable an adversary to run interactive commands and execute remote program functions. Our results show that by SCM we were able to identify distinct malware communities within the malware family, which revealed insights into the overall structure of the various binaries as well as possible temporal relationships between the binaries.

Deep UL2DL: Data-Driven Channel Knowledge Transfer From Uplink to Downlink

To remove the need for signaling overhead of feedback channels for transmitter channel state information (CSI) in Frequency Division Duplexing (FDD), we propose using convolutional neural networks and generative adversarial networks (GANs) to infer the downlink (DL) CSI by observing the uplink (UL) CSI. Our data-driven scheme exploits the fact that both DL and UL channels share the same propagation environment. As such, we extracted the environment information from UL channel response to a latent domain and then transferred the derived environment information from the latent domain to predict the DL channel. To prevent incorrect latent domain and the problem of oversimplistic assumptions, we did not use any specific parametric model and, instead, used data-driven approaches to discover the underlying structure of data without any prior model assumptions. To overcome the challenge of capturing the UL-DL joint distribution, we used a mean square error-based variant of the GAN structure with improved convergence properties called boundary equilibrium GAN. For training and testing we used simulated data of Extended Vehicular-A (EVA) and Extended Typical Urban (ETU) models. Simulation results verified that our methods can accurately infer and predict the downlink CSI from the uplink CSI for different multipath environments.

Nonlinear Multiview Analysis: Identifiability and Neural Network-Assisted Implementation

Multiview analysis aims at extracting shared latent components from data samples that are acquired in different domains, e.g., image, text, and audio. Classic multiview analysis, e.g., canonical correlation analysis (CCA), tackles this problem via matching the linearly transformed views in a certain latent domain. More recently, powerful nonlinear learning tools such as kernel methods and neural networks are utilized for enhancing the classic CCA. However, unlike linear CCA whose theoretical aspects are clearly understood, nonlinear CCA approaches are largely intuition-driven. In particular, it is unclear under what conditions the shared latent components across the views can be identified---while identifiability plays an essential role in many applications. In this work, we revisit nonlinear multiview analysis and address both the theoretical and computational aspects. Our work leverages a useful nonlinear model, namely, the post-nonlinear model, from the nonlinear mixture separation literature. Combining with multiview data, we take a nonlinear multiview mixture learning viewpoint, which is a natural extension of the classic generative models for linear CCA. From there, we derive a learning criterion. We show that minimizing this criterion leads to identification of the latent shared components up to certain ambiguities, under reasonable conditions. Our derivation and formulation also offer new insights and interpretations to existing deep neural network-based CCA formulations. On the computation side, we propose an effective algorithm with simple and scalable update rules. A series of simulations and real-data experiments corroborate our theoretical analysis.

Women's empowerment and child nutrition: The role of intrinsic agency.

Women's empowerment is associated with improved child nutrition, and both underpin the achievement of multiple Sustainable Development Goals (SDGs). We examined pathways by which women's empowerment influences child nutritional status. We pooled nationally representative data from Demographic and Health Surveys (2011–2016) collected from married women with children aged 6–24 months in Ethiopia, Kenya, Rwanda, Tanzania, and Uganda (n = 13,780). We operationalized child nutritional status using anemia, height-for-age z-score (HAZ), and weight-for-age z-score (WHZ). We operationalized women's empowerment using a validated measure comprised of three latent domains: social/human assets (“assets”), intrinsic agency (attitudes about intimate partner violence), and instrumental agency (influence in household decision making). We used structural equation models with latent constructs to estimate hypothesized pathways from women's empowerment to child nutritional status with further mediation by maternal body mass index (BMI) and stratification by wealth. Women's empowerment domains were directly and positively associated with maternal BMI (estimate±SE: assets, 0.17 ± 0.03; intrinsic agency, 0.23 ± 0.03; instrumental agency, 0.03 ± 0.01). Maternal BMI was directly and positively associated with child HAZ (0.08 ± 0.04) and child WHZ (0.35 ± 0.03). Assets were indirectly associated with child HAZ and WHZ through intrinsic agency and maternal BMI. In the lowest wealth category, the direct effects from women's empowerment to child nutritional status were significant (assets and instrumental agency were associated with anemia; intrinsic agency associated with HAZ). In the highest wealth category, direct effects from women's empowerment on child nutritional status were significant (intrinsic and instrumental agency associated with WHZ). Improving women's empowerment, especially intrinsic agency, in East Africa could improve child nutrition directly and via improved maternal nutrition. These findings suggest that efforts to realize SDG 5 may have spillover effects on other SDGs. However, strategies to improve nutrition through empowerment approaches may need to also address household resource constraints.

ANALYTICAL CONSIDERATIONS OF DEVELOPING A PHENOTYPIC AGING MEASURE: THE CONCEPTUAL FRAMEWORK MUST COME FIRST!

We propose a latent structural model framework where phenotypic aging is a latent variable influenced by chronological age, genes and environment. Within this framework, phenotypic age influences aging-related outcomes and is reflected by latent domains of aging (body composition, energetics, homeostasis, and neural functioning) reflected by biomarkers. First, we validate the framework by selecting age-associated domain-specific biomarkers and assessing internal consistency and convergent construct validity (Cronbach’s alpha). Using data from the Baltimore Longitudinal Study of Aging, within-domain Cronbach’s alphas ranged from 0.80 to 0.92, supporting convergent construct validity. Second, we evaluate two broad methods for combining biomarkers into one phenotypic age measure customized to different objectives: 1) confirmatory factor analysis of chronological age-adjusted biomarkers to create a measure to identify pleiotropic genetic and environmental mechanisms, and 2) machine-learning methods to create a measure optimizing predictive and concurrent criterion validity. This framework will enable evaluation of candidate biological mechanisms of aging.

Facilitating Calibration in High-Speed BCI Spellers via Leveraging Cross-Device Shared Latent Responses.

This paper proposes a novel device-to-device transfer-learning algorithm for reducing the calibration cost in a steady-state visual evoked potential (SSVEP)-based brain-computer interface (BCI) speller by leveraging electroencephalographic (EEG) data previously acquired by different EEG systems. The transferring is done by projecting the scalp-channel EEG signals onto a shared latent domain across devices. Three spatial filtering techniques, including channel averaging, canonical correlation analysis (CCA), and task-related component analysis (TRCA), were employed to extract the shared responses from different devices. The transferred data were integrated into a template-matching-based algorithm to detect SSVEPs. To evaluate its transferability, this paper conducted two sessions of simulated online BCI experiments with ten subjects using 40 visual stimuli modulated by joint frequency-phase coding method. In each session, two different EEG devices were used: first, the Quick-30 system (Cognionics, Inc.) with dry electrodes, and second, the ActiveTwo system (BioSemi, Inc.) with wet electrodes. The proposed method with CCA- and TRCA-based spatial filters achieved significantly higher classification accuracy compared with the calibration-free standard CCA-based method. This paper validated the feasibility and effectiveness of the proposed method in implementing calibration-free SSVEP-based BCIs. The proposed method has great potentials to enhance practicability and usability of real-world SSVEP-based BCI applications by leveraging user-specific data recorded in previous sessions even with different EEG systems and montages.

Fisher-discriminative regularized latent sparse transfer model

Top performing algorithms are trained on massive amounts of labeled data. Alternatively, domain adaptation (DA) provides an attractive way to address the few labeled tasks when the labeled data from a different but related domain are available. Motivated by Fisher criterion, we present the novel discriminative regularization term on the latent subspace which incorporates the latent sparse domain transfer (LSDT) model in a unified framework. The key underlying idea is to make samples from one class closer and farther away from different class samples. However, it is nontrivial to design the efficient optimization algorithm. Instead, we construct a convex surrogate relaxation optimization constraint to ease this issue by alternating direction method of multipliers (ADMM) algorithm. Subsequently, we generalize our model in the reproduced kernel Hilbert space (RKHS) for tracking the nonlinear domain shift. Empirical studies demonstrate the performance improvement on the benchmark vision dataset Caltech-4DA.

A Deep Collaborative Framework for Face Photo-Sketch Synthesis.

Great breakthroughs have been made in the accuracy and speed of face photo-sketch synthesis in recent years. Regression-based methods have gained increasing attention, which benefit from deeper and faster end-to-end convolutional neural networks. However, most of these models typically formulate the mapping from photo domain X to sketch domain Y as a unidirectional feedforward mapping, G: X → Y , and vice versa, F: Y → X ; thus, the utilization of mutual interaction between two opposite mappings is lacking. Therefore, we proposed a collaborative framework for face photo-sketch synthesis. The concept behind our model was that a middle latent domain ~Z between the photo domain X and the sketch domain Y can be learned during the learning procedure of G: X → Y and F: Y → X by introducing a collaborative loss that makes full use of two opposite mappings. This strategy can constrain the two opposite mappings and make them more symmetrical, thus making the network more suitable for the photo-sketch synthesis task and obtaining higher quality generated images. Qualitative and quantitative experiments demonstrated the superior performance of our model in comparison with the existing state-of-the-art solutions.