The development of gaze following in monolingual and bilingual infants: A multi-lab study

Determining the meanings of words requires language learners to attend to what other people say. However, it behooves a young language learner to simultaneously attend to what other people attend to, for example, by following the direction of their eye gaze. Sensitivity to cues such as eye gaze might be particularly important for bilingual infants, as they encounter less consistency between words and objects than monolinguals, and do not always have access to the same word learning heuristics (e.g., mutual exclusivity). In a pre-registered study, we tested the hypothesis that bilingual experience would lead to a more pronounced ability to follow another’s gaze. We used the gaze-following paradigm developed by Senju & Csibra (2008) to test a total of 93 6–9 month-old and 229 12–15 month-old monolingual and bilingual infants, in 11 labs located in 8 countries. Monolingual and bilingual infants showed similar gaze-following abilities, and both groups showed age-related improvements in speed, accuracy, frequency and duration of fixations to congruent objects. Unexpectedly, bilinguals tended to make more frequent fixations to onscreen objects, whether or not they were cued by the actor. These results suggest that gaze sensitivity is a fundamental aspect of development that is robust to variation in language exposure.

Merging Weak and Active Supervision for Semantic Parsing

A semantic parser maps natural language commands (NLs) from the users to executable meaning representations (MRs), which are later executed in certain environment to obtain user-desired results. The fully-supervised training of such parser requires NL/MR pairs, annotated by domain experts, which makes them expensive to collect. However, weakly-supervised semantic parsers are learnt only from pairs of NL and expected execution results, leaving the MRs latent. While weak supervision is cheaper to acquire, learning from this input poses difficulties. It demands that parsers search a large space with a very weak learning signal and it is hard to avoid spurious MRs that achieve the correct answer in the wrong way. These factors lead to a performance gap between parsers trained in weakly- and fully-supervised setting. To bridge this gap, we examine the intersection between weak supervision and active learning, which allows the learner to actively select examples and query for manual annotations as extra supervision to improve the model trained under weak supervision. We study different active learning heuristics for selecting examples to query, and various forms of extra supervision for such queries. We evaluate the effectiveness of our method on two different datasets. Experiments on the WikiSQL show that by annotating only 1.8% of examples, we improve over a state-of-the-art weakly-supervised baseline by 6.4%, achieving an accuracy of 79.0%, which is only 1.3% away from the model trained with full supervision. Experiments on WikiTableQuestions with human annotators show that our method can improve the performance with only 100 active queries, especially for weakly-supervised parsers learnt from a cold start. 1

Memetic algorithm for short messaging service spam filter using text normalization and semantic approach

Today’s popularity of the short messages services (SMS) has created a propitious environment for spamming to thrive. Spams are unsolicited advertising, adult-themed or inappropriate content, premium fraud, smishing and malware. They are a constant reminder of the need for an effective spam filter. However, SMS limitations of 160-charcaters and 140-bytes size as well as its being rippled with slangs, emoticons and abbreviations further inhibits effective training of models to aid accurate classification. The study proposes Genetic Algorithm Trained Bayesian Network solution that seeks to normalize noisy feats, expand text via use of lexicographic and semantic dictionaries that uses word sense disambiguation technique to train the underlying learning heuristics. And in turn, effectively help to classify SMS in spam and legitimate classes. Hybrid model comprises of text preprocessing, feature selection as well as training and classification section. Study uses a hybrid Genetic Algorithm trained Bayesian model for which the GA is used for feature selection; while, the Bayesian algorithm is used as classifier.

Design and evaluation of practice-oriented materials fostering students’ development of problem-solving competence

In a design-research project on problem-solving, theory-based and practice-oriented materials were developed with the goal of fostering systematical development of students’ problem-solving competence in a targeted manner by learning heuristics. Special attention was given to working backward strategy, which has been shown difficult for students to learn and use. In the study, 14 Grade 5 students participated in explicit heuristic training. The results show that even though the students intuitively reversed their thought processes before the explicit training, they experienced difficulties when solving complex reversing tasks, which improved considerably after explicit heuristic training. Thus, the study results showed that the developed materials using design-based research-approach promoted the development of students’ flexibility of thought when problem-solving by working backward. At the end of the paper, the results are discussed with regard to their theoretical and practical implications.