Kurdish Adaptation of Arabic Loan Consonants

This paper addresses the status of the Arabic loan consonants in Central Kurdish (CK). Based on the Arabic loanwords, it assesses different scenarios on how the foreign consonants are adapted. The paper finds out that Arabic loan consonants in CK can be classified into three groups: Consonants that are part of the phonemic inventory of both languages; consonants that are borrowed faithfully, i.e. without adaptation and finally consonants that are not allowed in the phonemic inventory of CK, i.e. require feature adaptation. The paper also makes contribution to the theories of loan adaptation. It shows that neither Phonological Stance Model nor Phonetic Stance Model can account for the way Arabic consonants are (un)adapted in CK. The faithful borrowing of guttural consonants and the adaptation of dental fricatives and emphatics to match the phonemic inventory of CK shows that there are active marking statements that (dis)allow a combination of features that form a segment. Some other factors also play roles in the faithful borrowing of the loan consonants such as frequency of the loanwords with loan phonemes, orthographic input and the sensitivity of the faithful pronunciation of the loanwords such as the loanwords that are proper names. Common proper names with guttural phonemes are borrowed faithfully.

Impaired Adaptation and Laminar Processing of the Oddball Paradigm in the Primary Visual Cortex of Fmr1 KO Mouse

Both adaptation and novelty detection are an integral part of sensory processing. Recent animal oddball studies have advanced our understanding of circuitry underlying contextual processing in early sensory areas. However, it is unclear how adaptation and mismatch (MM) responses depend on the tuning properties of neurons and their laminar position. Furthermore, given that reduced habituation and sensory overload are among the hallmarks of altered sensory perception in autism, we investigated how oddball processing might be altered in a mouse model of fragile X syndrome (FX). Using silicon probe recordings and a novel spatial frequency (SF) oddball paradigm, we discovered that FX mice show reduced adaptation and enhanced MM responses compared to control animals. Specifically, we found that adaptation is primarily restricted to neurons with preferred oddball SF in FX compared to WT mice. Mismatch responses, on the other hand, are enriched in the superficial layers of WT animals but are present throughout lamina in FX animals. Last, we observed altered neural dynamics in FX mice in response to stimulus omissions. Taken together, we demonstrated that reduced feature adaptation coexists with impaired laminar processing of oddball responses, which might contribute to altered sensory perception in FX syndrome and autism.

Unsupervised Adversarial Domain Adaptation with Error-Correcting Boundaries and Feature Adaption Metric for Remote-Sensing Scene Classification

Unsupervised domain adaptation (UDA) based on adversarial learning for remote-sensing scene classification has become a research hotspot because of the need to alleviating the lack of annotated training data. Existing methods train classifiers according to their ability to distinguish features from source or target domains. However, they suffer from the following two limitations: (1) the classifier is trained on source samples and forms a source-domain-specific boundary, which ignores features from the target domain and (2) semantically meaningful features are merely built from the adversary of a generator and a discriminator, which ignore selecting the domain invariant features. These issues limit the distribution matching performance of source and target domains, since each domain has its distinctive characteristic. To resolve these issues, we propose a framework with error-correcting boundaries and feature adaptation metric. Specifically, we design an error-correcting boundaries mechanism to build target-domain-specific classifier boundaries via multi-classifiers and error-correcting discrepancy loss, which significantly distinguish target samples and reduce their distinguished uncertainty. Then, we employ a feature adaptation metric structure to enhance the adaptation of ambiguous features via shallow layers of the backbone convolutional neural network and alignment loss, which automatically learns domain invariant features. The experimental results on four public datasets outperform other UDA methods of remote-sensing scene classification.