UrduAI: Writeprints for Urdu Authorship Identification

The authorship identification task aims at identifying the original author of an anonymous text sample from a set of candidate authors. It has several application domains such as digital text forensics and information retrieval. These application domains are not limited to a specific language. However, most of the authorship identification studies are focused on English and limited attention has been paid to Urdu. However, existing Urdu authorship identification solutions drop accuracy as the number of training samples per candidate author reduces and when the number of candidate authors increases. Consequently, these solutions are inapplicable to real-world cases. Moreover, due to the unavailability of reliable POS taggers or sentence segmenters, all existing authorship identification studies on Urdu text are limited to the word n-grams features only. To overcome these limitations, we formulate a stylometric feature space, which is not limited to the word n-grams feature only. Based on this feature space, we use an authorship identification solution that transforms each text sample into a point set, retrieves candidate text samples, and relies on the nearest neighbors classifier to predict the original author of the anonymous text sample. To evaluate our solution, we create a significantly larger corpus than existing studies and conduct several experimental studies that show that our solution can overcome the limitations of existing studies and report an accuracy level of 94.03%, which is higher than all previous authorship identification works.

Attention-Based Deep Multiple-Instance Learning for Classifying Circular RNA and Other Long Non-Coding RNA

Circular RNA (circRNA) is a distinguishable circular formed long non-coding RNA (lncRNA), which has specific roles in transcriptional regulation, multiple biological processes. The identification of circRNA from other lncRNA is necessary for relevant research. In this study, we designed attention-based multi-instance learning (MIL) network architecture fed with a raw sequence, to learn the sparse features of RNA sequences and to accomplish the circRNAs identification task. The model outperformed the state-of-art models. Moreover, following the validation of the attention mechanism effectiveness by the handwritten digit dataset, the key sequence loci underlying circRNA’s recognition were obtained based on the corresponding attention score. Then, motif enrichment analysis identified some of the key motifs for circRNA formation. In conclusion, we designed deep learning network architecture suitable for learning gene sequences with sparse features and implemented it for the circRNA identification task, and the model has strong representation capability in the indication of some key loci.

Power mediates the processing of gender during sibilant categorization

Prior studies suggest that listeners are more likely to categorize a sibilant ranging acoustically from [∫] to [s] as /s/ if provided auditory or visual information about the speaker that suggests male gender. Social cognition can also be affected by experimentally induced differences in power. A powerful individual’s impression of another tends to show greater consistency with the other person’s broad social category, while a powerless individual’s impression is more consistent with the specific pieces of information provided about the other person. This study investigated whether sibilant categorization would be influenced by power when the listener is presented with inconsistent sources of information about speaker gender. Participants were experimentally primed for behavior consistent with powerful or powerless individuals. They then completed a forced choice identification task: They saw a visual stimulus (a male or female face) and categorized an auditory stimulus (ranging from ‘shy’ to ‘sigh’) as /∫/ or /s/. As expected, participants primed for high power were sensitive to a single cue to gender, while those who received the low power prime were sensitive to both, even if the cues did not match. This result suggests that variability in listener power may cause systematic differences in phonetic perception.

Effects of repeating after stimuli sounds during computer assisted HVPT on Japanese learners’ perception and production of English fricatives

This study investigated the effects of using an online High Variability Phonetic Training (HVPT) program for university students whose first language is Japanese and who studied English as a foreign language. The target sounds were English fricatives, which many of the participants felt were challenging to distinguish and articulate. The training was conducted for five weeks; each week consisted of three HVPT sessions. The training process consisted of an initial explanation of how to articulate the sounds followed by a video that showed the physical articulative movements. The participants listened to stimuli sounds, produced the sounds after each stimulus, and completed an identification task. Based on pre and posttests, the perception performance of the participants improved significantly through the HVPT training. Regarding production, a positive effect was found for one phoneme, while no difference was found for the other. Questionnaire results were generally positive but indicated the need for explicit instruction for some of the fricatives that exhibited relatively lower successful production and perception rates.

Anger, race, and the neurocognition of threat: attention, inhibition, and error processing during a weapon identification task

This study measured event-related brain potentials (ERPs) to test competing hypotheses regarding the effects of anger and race on early visual processing (N1, P2, and N2) and error recognition (ERN and Pe) during a sequentially primed weapon identification task. The first hypothesis was that anger would impair weapon identification in a biased manner by increasing attention and vigilance to, and decreasing recognition and inhibition of weapon identification errors following, task-irrelevant Black (compared to White) faces. Our competing hypothesis was that anger would facilitate weapon identification by directing attention toward task-relevant stimuli (i.e., objects) and away from task-irrelevant stimuli (i.e., race), and increasing recognition and inhibition of biased errors. Results partially supported the second hypothesis, in that anger increased early attention to faces but minimized attentional processing of race, and did not affect error recognition. Specifically, angry (vs. neutral) participants showed increased N1 to both Black and White faces, ablated P2 race effects, and topographically restricted N2 race effects. Additionally, ERN amplitude was unaffected by emotion, race, or object type. However, Pe amplitude was affected by object type (but not emotion or race), such that Pe amplitude was larger after the misidentification of harmless objects as weapons. Finally, anger slowed overall task performance, especially the correct identification of harmless objects, but did not impact task accuracy. Task performance speed and accuracy were unaffected by the race of the face prime. Implications are discussed.

How explicit instruction improves phonological awareness and perception of L2 sound contrasts in younger and older adults

Despite the importance of conscious awareness in second language acquisition theories, little is known about how L2 speech perception can be improved by explicit phonetic instruction. This study examined the relationship between phonological awareness and perception in Dutch younger and older adult L2 listeners, focusing on English contrasts of two types: a familiar contrast in an unfamiliar position (word-final /t/-/d/) and an unfamiliar contrast (/æ/-/ε/). Awareness was assessed with a task in which written minimal pairs and homophone pairs had to be judged as sounding the same or different. Perception was assessed with a two-alternative forced-choice identification task with auditorily presented words from minimal pairs. We investigated whether listeners’ awareness and perception improved after a video-based explicit instruction that oriented their attention to one of these contrasts, and we tested whether including information about the phonetic cue of vowel duration increased learning. Awareness and perception of each contrast were shown to be moderately correlated at the study’s outset. Furthermore, awareness and perception for each contrast generally improved more after the instruction drawing attention to that contrast. However, the effectiveness of explicit phonetic instruction varied depending on the combination of the contrast, cue information, and listener age group.

Function shaping in deep learning

This work describes the importance of loss functions and related methods for deep reinforcement learning and deep metric learning. A novel MDQN loss function outperformed DDQN loss function in PLE computer game environments, and a novel Exponential Triplet loss function outperformed the Triplet loss function in the face re-identification task with VGGFace2 dataset reaching 85,7 % accuracy using zero-shot setting. This work also presents a novel UNet-RNN-Skip model to improve the performance of the value function for path planning tasks.

Floating Boundary Tone: Production and Perception of Syntactically Unmarked Polar Question in Tianjin Mandarin

The present study investigates the intonational tune of syntactically unmarked polar question in Tianjin Mandarin. A production study was conducted to examine the phonological features of the syntactically unmarked polar question (a.k.a intonational yes/no question) tune by comparing against the statement tune. The results show a significant register lift HR and a high floating boundary tone H̥I. The tone shape and tone register played a significant role in how the tunes vary. A tune identification task then further verifies whether the two prosodic features concluded from the production are used in perception. The results showed that both the register difference and the boundary tone made a difference in native speakers' perception in discriminating questions from statements.

Evaluation of Intervention Effectiveness of Sensory Compensatory Training with Tactile Discrimination Feedback on Sensorimotor Dysfunction of the Hand after Stroke

We investigated the intervention effect of training using a feedback-type tactile discrimination system on sensorimotor dysfunction of the hand after a stroke. A human male subject with sensorimotor dysfunction in his left hand after a stroke was asked to perform peg manipulation practice, a building block stacking task, and a material identification task for 10 min each for six weeks. During the activities, a tactile discrimination feedback system was used. The system is a device that detects the vibration information generated when touching an object with a hand and that feeds back the captured information in real time as vibration information. After the intervention, in addition to the reorganization of the sensorimotor areas, the deep sensation, sense of agency, numbness, amount of use, and quality of the left-hand movement improved. Our results suggest that training with the use of a feedback system could be a new form of rehabilitation for sensorimotor dysfunction of the hand.

Unmasking the Effects of Orthography, Semantics, and Phonology on 2AFC Visual Word Perceptual Identification

In the two-alternative forced-choice (2AFC) task, the target stimulus is presented very briefly, and the participants must choose between two options as to which was the presented target. Some past research (Grossi et al., 2009; Haro et al., 2019) has assumed that the 2AFC word identification task isolates orthographic effects, despite orthographic, semantic, and phonological differences between the alternative options. If so, performance should not differ between word target/nonword foil pairs and British/American word pairs, the latter of which only differ orthographically. In Experiment 1, accuracy and sensitivity were higher during word/nonword trials than British/American trials when participants stated their response was not a guess, demonstrating that phonological/semantic processing contributes to 2AFC performance. In Experiment 2, target visibility was manipulated by increasing the contrast between target and mask for half the trials. Experiment 2 showed that target visibility did not interact with pair type on reaction time, which suggests phonological/semantic processing did not result in feedback to orthographic encoding in this task. This study demonstrates the influence of phonological/semantic processing on word perceptual identification, and shows that 2AFC word identification does not isolate orthographic effects when word/nonword pairs are used, but using British/American word pairs provides a method for doing so. Implications for models and future research are discussed.