Search

This study investigates the effects of prior intentions and intentions in action during the initial stage of a process of design in search of correlations between these cognitive processes and the quality of the design outcomes. We found that the quality of the outcomes was not influenced by sketching. Nevertheless, most of the participants said that they preferred sketching, and their verbal reports can be seen as indicative of the formation of prior intentions. Therefore, we argue that prior intentions have an important role even within a process of pure material engagement. This has important implications for understanding the active role of human agency and our capacity to form prior intentions with pragmatic effects in the world. [ABSTRACT FROM AUTHOR]

Students with hearing loss (HL) often fall behind hearing peers in complex language tasks such as narrative writing. This study explored the effects of school grade, gender, cognitive and linguistic predisposition and audiological factors on narrative text quality in this target group. Eleven students with HL in Grades 5–6 and 7–8 (age 12–15) who took part in a writing intervention wrote four narrative texts over six months. A trained panel rated text quality. The effects of the students' working memory capacity, language comprehension, reading comprehension, school grade and gender and the intervention were analyzed as a mixed-effects regression model. Audiological factors were considered separately. The analysis showed that throughout the period, texts written by female students in Grade 7–8 received the highest text quality ratings, while those written by male students in Grade 7–8 received the lowest ratings. There was no effect of the intervention, or of the linguistic and cognitive measures. The students with the lowest text quality ratings received amplification later than those with high ratings, but HL severity was not associated with text quality. Hearing loss severity was not a decisive factor in narrative text quality. The intervention which the students took part in is potentially effective, with some adaptation to the special needs of students with HL. The strong gender effects are discussed. [ABSTRACT FROM AUTHOR]

• We study feature replay with compressed exemplars for online continual learning. • We show the importance of replaying features from intermediate layers of the network. • We propose ACAE-REMIND enabling better feature compression based on auto-encoders. • The method achieves good performance in several online continual learning settings. • Our online method even surpasses several offline continual learning methods. Online continual learning aims to learn from a non-IID stream of data from a number of different tasks, where the learner is only allowed to consider data once. Methods are typically allowed to use a limited buffer to store some of the images in the stream. Recently, it was found that feature replay, where an intermediate layer representation of the image is stored (or generated) leads to superior results than image replay, while requiring less memory. Quantized exemplars can further reduce the memory usage. However, a drawback of these methods is that they use a fixed (or very intransigent) backbone network. This significantly limits the learning of representations that can discriminate between all tasks. To address this problem, we propose an auxiliary classifier auto-encoder (ACAE) module for feature replay at intermediate layers with high compression rates. The reduced memory footprint per image allows us to save more exemplars for replay. In our experiments, we conduct task-agnostic evaluation under online continual learning setting and get state-of-the-art performance on ImageNet-Subset, CIFAR100 and CIFAR10 dataset. [ABSTRACT FROM AUTHOR]

In this study, we investigated the prosody of domestic cat meows produced in different contexts. Prosodic cues – i.e., variation in intonation, duration, voice quality and fundamental frequency – in humans as well as in nonhuman animals carry information about idiosyncratic traits of the signaller, including sex, age, and physical and mental state. The duration, fundamental frequency (F0) and intonation in a sample of 969 meows recorded in seven different contexts (i.e., cuddle , door , food , greeting , lifting , play , cat carrier) were analysed using linear mixed effects regression and generalized additive models. In this, we controlled for cat age and sex, as meows produced by old cats had lower mean F0 than those produced by young cats, and female cats produced meows with higher mean F0 than male cats. We found significant effects of context on duration and mean F0, but not on F0 range. Furthermore, the results showed that the intonation of meows produced by cats in a cat carrier displayed a falling pattern, while that of meows produced in cuddle and door contexts was relatively level, and that of meows produced in the other contexts consisted of combinations of rising and falling. The average slope of meows produced in cat carrier and play contexts was negative, while that of meows produced in the other contexts was positive. We argue that this prosodic variation reflects the cats' mental or emotional state, because of valence and arousal differences associated with the various contexts that were included in the study. Further studies will need to confirm this. In addition, we also plan additional analyses of spectral and voice quality parameters in meows and other cat vocalisation types. • The vocal prosody of domestic cat meows varies with the physical context. • Domestic cat meows vary in duration, mean fundamental frequency (F0) and intonation. • Young and female cat meows have higher F0 than old and male cat meows. • Meows in cat carriers have falling intonation, but is more varied in other contexts. [ABSTRACT FROM AUTHOR]

Observational learning has shown to be a successful intervention for writing. Until now, however, studies have only been performed with normal-hearing participants, usually high school or university students. Additionally, there have been conflicting results in whether subjective text quality correlates with one or more objectively measured text characteristics. In this study, we measured the effect of observational learning in a group of four university students with hearing impairment, and compared the results with those of a group of 10 students with normal hearing who did the same intervention, and those of a control group consisting of 10 students with normal hearing who did not do the intervention. Subjective text quality ratings and nine objectively measured text characteristics were collected for three argumentative texts written by each of the participants. In between writing these three texts, the participants in the experimental groups watched a video of a model writer who read out loud and corrected a similar kind of text. The statistical analysis showed significant correlations between the subjective ratings and four out of the nine objective measures, but no significant intervention effect. These findings suggest that observation-learning intervention is most effective when the model writer is a peer learner, and when the intervention is stretched out over time. Additionally, the method may be better suited for learners younger than the ones who were included in the present study. [ABSTRACT FROM AUTHOR]

Abstract In linguistics and in the philosophy of language it is standardly assumed that know is a factive verb, meaning that a sentence such as X knows that p , when uttered in its positive declarative form, presupposes, in fact entails, the truth of its complement. A problem for this analysis is the fact that the verb know can be used non-factively in contexts where it is evident that the proposition expressed by the subordinate clause is not true. In order to account for non-factive uses of know , two main solutions have been advanced in the literature. Hazlett (2009, 2010, 2012) proposes that know is not semantically factive and a sentence such as X knows that p does not entail, but only pragmatically implies p. On the other hand, Tsohatzidis (2012) argues that know is lexically ambiguous between a factive and a non-factive sense: when know is used in its factive sense, a sentence such as X knows that p entails p , whereas, when know occurs in its non-factive sense, it does not. As shown in recent works by Colonna Dahlman (2015, 2016, 2017b), the phenomenon at issue—the possibility for a speaker to use know in cases where the proposition expressed by the clause embedded under 'knows' is not true—is not unique to English, but occurs, for instance, also in Italian. We carried out a Truth Judgment Task to test the hypothesis that the Italian lexical item 'sa' (' knows ') is ambiguous. Our findings are consistent with the lexical ambiguity hypothesis, and cannot be explained by Hazlett's pragmatic solution. Highlights • Italian sapere is lexically ambiguous between a factive and a non-factive sense. • The results of a Truth Judgment Task confirm the lexical ambiguity hypothesis. • Our findings are not compatible with the predictions outlined by a pragmatic view. [ABSTRACT FROM AUTHOR]

Most approaches to human attribute and action recognition in still images are based on image representation in which multi-scale local features are pooled across scale into a single, scale-invariant encoding. Both in bag-of-words and the recently popular representations based on convolutional neural networks, local features are computed at multiple scales. However, these multi-scale convolutional features are pooled into a single scale-invariant representation. We argue that entirely scale-invariant image representations are sub-optimal and investigate approaches to scale coding within a bag of deep features framework. Our approach encodes multi-scale information explicitly during the image encoding stage. We propose two strategies to encode multi-scale information explicitly in the final image representation. We validate our two scale coding techniques on five datasets: Willow, PASCAL VOC 2010, PASCAL VOC 2012, Stanford-40 and Human Attributes (HAT-27). On all datasets, the proposed scale coding approaches outperform both the scale-invariant method and the standard deep features of the same network. Further, combining our scale coding approaches with standard deep features leads to consistent improvement over the state of the art. [ABSTRACT FROM AUTHOR]

Previous work in the areas of organizational trust repair and crisis communication has provided conflicting answers to the question of whether denial can be more effective than apology in repairing stakeholder trust in a company following an integrity-based violation. This article reports the results of an experiment designed to (i) test the effects of these two strategies on individuals’ trust in a company accused of corruption, and (ii) determine whether and how evidence of the company’s guilt influences stakeholder reactions to its trust repair message. The results demonstrate that, when evidence against the company is weak, trust is restored more successfully with a denial than an apology. Contrary to our hypothesis, denial was found to outperform apology in repairing perceptions of the company’s integrity and benevolence even in the face of strong evidence, and it was as effective as apology in restoring perceived ability and trusting intentions. These results provide empirical evidence for the ‘paradoxical effect’ that an open and honest attitude can, in the short term, be more detrimental to organizations than a defensive strategy. More research on the factors that determine the credibility and persuasiveness of corporate denial is called for. [ABSTRACT FROM AUTHOR]

In the last few years, the traditional analysis of know as a factive verb has been lively debated by linguists and philosophers of language: several scholars have pointed out that know may be used non-factively in ordinary language. The aim of the present study is to expand this inquiry to other cognitive factive verbs than know , such as discover , realize , etc., and to investigate cross-linguistically the question of whether know and other cognitive factive verbs may occur in non-factive contexts, that is, in contexts where it is clear that the embedded proposition is false. Moreover, we investigate whether so-called evidential uses of cognitive factive verbs are acceptable across languages. We administered an online survey to native speakers of nine different languages (English, French, German, Greek, Italian, Portuguese, Serbian, Spanish, and Swedish), and we found considerable cross-linguistic variation in the acceptability of the use of know and other cognitive factive verbs in non-factive contexts. For Italian and English, we put forward the claim that non-factive uses of cognitive factives instantiate a case of polysemy resulting from a process of semantic change that moves along a three-step pattern: from a factive sense to a more general non-factive sense to a non-factive sense characterized by an evidential function. • Cognitive factive verbs are investigated cross-linguistically. • Know and other cognitive factive verbs are used non-factively in ordinary language. • Cognitive factive verbs may be used to fulfill an evidential strategy. • An Acceptability Judgment Task was submitted to native speakers of nine languages. • Evidential uses may be a step in a process of defactivization of factive verbs. [ABSTRACT FROM AUTHOR]

Highlights: [•] Canonical antonyms give rise to significantly lower N400 amplitudes than non-canonical antonyms or unrelated word pairs. [•] Without context, the difference in N400 amplitude for non-canonical antonyms and unrelated word pairs is not significant. [•] With context, the difference in N400 amplitude between non-canonical antonyms and unrelated word pairs is significant. [ABSTRACT FROM AUTHOR]

Color names are required in real-world applications such as image retrieval and image annotation. Traditionally, they are learned from a collection of labeled color chips. These color chips are labeled with color names within a well-defined experimental setup by human test subjects. However, naming colors in real-world images differs significantly from this experimental setting. In this paper, we investigate how color names learned from color chips compare to color names learned from real-world images. To avoid hand labeling real-world images with color names, we use Google Image to collect a data set. Due to the limitations of Google Image, this data set contains a substantial quantity of wrongly labeled data. We propose several variants of the PLSA model to learn color names from this noisy data. Experimental results show that color names learned from real-world images significantly outperform color names learned from labeled color chips for both image retrieval and image annotation. [ABSTRACT FROM AUTHOR]

Color constancy is the ability to measure colors of objects independent of the color of the light source. A well-known color constancy method is based on the gray-world assumption which assumes that the average reflectance of surfaces in the world is achromatic. In this paper, we propose a new hypothesis for color constancy namely the gray-edge hypothesis, which assumes that the average edge difference in a scene is achromatic. Based on this hypothesis, we propose an algorithm for color constancy. Contrary to existing color constancy algorithms, which are computed from the zero-order structure of images, our method is based on the derivative structure of images. Furthermore, we propose a framework which unifies a variety of known (gray-world, max-RGB, Minkowski norm) and the newly proposed gray-edge and higher order gray-edge algorithms. The quality of the various instantiations of the framework is tested and compared to the state-of-the-art color constancy methods on two large data sets of images recording objects under a large number of different light sources. The experiments show that the proposed color constancy algorithms obtain comparable results as the state-of-the-art color constancy methods with the merit of being computationally more efficient. [ABSTRACT FROM AUTHOR]

Abstract: Speech variation of speakers is a crucial issue for speaker recognition and identification, especially for forensic practice. Greater intra-speaker variation is one main reason for incorrect speaker identification in real forensic situations. Understanding the stability of acoustic parameters and their variation in speech is therefore significant for the evaluation of effective parameters for speaker identification. In this paper, all vowels in Standard Chinese including five monophthongs, eight diphthongs and four triphthongs were combined with lateral /l/. Finally, 15 lateral syllables with different tones for 10 speakers were selected and acoustically analyzed. Central frequencies of the first four formants for each syllable were measured for quantitative comparison of intra- and inter-speaker variation in order to provide a general idea of speaker variation in Standard Chinese, and finally serving for forensic application. Results show that the overall intra-speaker variation is less than the inter-speaker variation in great extent under laboratory condition though occasionally they are contrary. This supports the basis for forensic speaker identification, that is, intra-speaker variation should be less than inter-speaker variation in many acoustic features, and further validates the probability and reliability of forensic speaker identification. [Copyright &y& Elsevier]

The aim of salient feature detection is to find distinctive local events in images. Salient features are generally determined from the local differential structure of images. They focus on the shape-saliency of the local neighborhood. The majority of these detectors are luminance-based, which has the disadvantage that the distinctiveness of the local color information is completely ignored in determining salient image features. To fully exploit the possibilities of salient point detection in color images, color distinctiveness should be taken into account in addition to shape distinctiveness. In this paper, color distinctiveness is explicitly incorporated into the design of saliency detection. The algorithm, called color saliency boosting, is based on an analysis of the statistics of color image derivatives. Color saliency boosting is designed as a generic method easily adaptable to existing feature detectors. Results show that substantial improvements in information content are acquired by targeting color salient features. [ABSTRACT FROM AUTHOR]

Luminance-based features are widely used as low-level input for computer vision applications, even when color data is available. The extension of feature detection to the color domain prevents information loss due to isoluminance and allows us to exploit the photometric information. To fully exploit the extra information in the color data, the vector nature of color data has to be taken into account and a sound framework is needed to combine feature and photometric invariance theory. In this paper, we focus on the structure tensor, or color tensor, which adequately handles the vector nature of color images. Further, we combine the features based on the color tensor with photometric invariant derivatives to arrive at photometric invariant features. We circum- vent the drawback of unstable photometric invariants by deriving an uncertainty measure to accompany the photometric invariant derivatives. The uncertainty is incorporated in the color tensor, hereby allowing the computation of robust photometric invariant features. The combination of the photometric invariance theory and tensor-based features allows for detection of a variety of features such as photometric invariant edges, corners, optical flow, and curvature. The proposed features are tested for noise char- acteristics and robustness to photometric changes. Experiments show that the proposed features are robust to scene incidental events and that the proposed uncertainty measure improves the applicability of full invariants. [ABSTRACT FROM AUTHOR]

Feature detection is used in many computer vision applications such as image segmentation, object recognition, and image retrieval. For these applications, robustness with respect to shadows, shading, and specularities is desired. Features based on derivatives of photometric invariants, which we will call full in variants, provide the desired robustness. However, because computation of photometric invariants involves nonlinear transformations, these features are unstable and, therefore, impractical for many applications. We propose a new class of derivatives which we refer to as quasi-invariants. These quasi-invariants are derivatives which share with full photometric invariants the property that they are insensitive for certain photometric edges, such as shadows or specular edges, but without the inherent instabilities of full photometric invariants. Experiments show that the quasi-invariant derivatives are less sensitive to noise and introduce less edge displacement than full invariant derivatives. Moreover, quasi-invariants significantly outperform the full invariant derivatives in terms of discriminative power. [ABSTRACT FROM AUTHOR]

Presents a study which measured the amount of local gradient energy along the model gradient vector fields as a function of model curvature using parameterized curvilinear models. Methodology used; Results; Conclusions.

Would metaphors used in the context of psychotherapy by people who experience various forms of anxiety disorders differ from those used by people who experience stress? We investigated this question with the help of the Motivation & Sedimentation Model (MSM), a theory of meaning-making developed within the synthetic new discipline of cognitive semiotics. The analysis of a sample of ten transcripts of psychotherapy sessions concerning the topic of anxiety, and a comparable sample concerning stress, showed a significantly stronger proportion of conventionalized metaphors in the stress sample, and a marginally significant difference in the number of innovative metaphors in the anxiety sample. These results suggest that lived experience of an anxiety disorder or another form of maladaptive anxiety affects metaphorical meaning-making, and manifests itself in spontaneous metaphor use. Furthermore, as a result of the conceptual and the empirical investigations of the topic, we propose novel theoretical and operational definitions of the notion of metaphoricity. [ABSTRACT FROM AUTHOR]

Gender inflection for animated nouns in Italian presents challenges influenced by societal pressures and linguistic structure, especially in morphologically complex words like compounds. The study investigates gender inflection distribution in compounds with capo- compared to other nouns (i.e., occupations traditionally performed by women, by men, and the word capo in isolation), exploring the interplay of social, etymological and morphological factors. 192 native Italian speakers inflected masculine nouns to feminine forms after hearing the stimulus. Results reveal that respondents’ attitudes towards gender-fair language significantly determine the use of feminine, indicating a complex interplay between linguistic structures and social perceptions. Despite historical resistance, the word capa in isolation shows increasing acceptance, challenging entrenched norms. In compounds, capo- element’s gender inflection appears more resistant due to morphological complexity, with an interaction with number. This study advances our understanding of gender inflection, with implications for broader conversations about gender representation and language inclusivity. [ABSTRACT FROM AUTHOR]

On-board vision systems may need to increase the number of classes that can be recognized in a relatively short period. For instance, a traffic sign recognition system may suddenly be required to recognize new signs. Since collecting and annotating samples of such new classes may need more time than we wish, especially for uncommon signs, we propose a method to generate these samples by combining synthetic images and Generative Adversarial Network (GAN) technology. In particular, the GAN is trained on synthetic and real-world samples from known classes to perform synthetic-to-real domain adaptation, but applied to synthetic samples of the new classes. Using the Tsinghua dataset with a synthetic counterpart, SYNTHIA-TS, we have run an extensive set of experiments. The results show that the proposed method is indeed effective, provided that we use a proper Convolutional Neural Network (CNN) to perform the traffic sign recognition (classification) task as well as a proper GAN to transform the synthetic images. Here, a ResNet101-based classifier and domain adaptation based on CycleGAN performed extremely well for a ratio ∼ 1 / 4 for new/known classes; even for more challenging ratios such as ∼ 4 / 1 , the results are also very positive. [ABSTRACT FROM AUTHOR]

• First language attrition affects Italian null pronouns. • Late bilinguals select the closest available antecedent (i.e., the object). • First language attrition affects processing. This study investigates whether L1 attrition effects on anaphora resolution decrease with L1 re-immersion. A group of 20 Italian-Swedish late bilinguals was tested once before and once after their summer vacation in Italy, and compared with a control group of 21 Italian monolinguals that was also tested once before and once after a similar time interval. Both groups assigned overt and null pronouns to a subject or an object antecedent in a self-paced comprehension task. The results of the study suggest that attrition affects null pronouns, an outcome inconsistent with previous findings wherein attrition effects were limited to Italian overt pronouns (e.g., Tsimpli et al., 2004). This outcome supports the hypothesis that attrition affects processing rather than representations. Moreover, in the second session, both groups improved in terms of antecedent assignment and response times. This finding suggests a task effect rather than a re-immersion effect. [ABSTRACT FROM AUTHOR]

Given the often enormous effort required to train GANs, both computationally as well as in dataset collection, the re-use of pretrained GANs largely increases the potential impact of generative models. Therefore, we propose a novel knowledge transfer method for generative models based on mining the knowledge that is most beneficial to a specific target domain, either from a single or multiple pretrained GANs. This is done using a miner network that identifies which part of the generative distribution of each pretrained GAN outputs samples closest to the target domain. Mining effectively steers GAN sampling towards suitable regions of the latent space, which facilitates the posterior finetuning and avoids pathologies of other methods, such as mode collapse and lack of flexibility. Furthermore, to prevent overfitting on small target domains, we introduce sparse subnetwork selection, that restricts the set of trainable neurons to those that are relevant for the target dataset. We perform comprehensive experiments on several challenging datasets using various GAN architectures (BigGAN, Progressive GAN, and StyleGAN) and show that the proposed method, called MineGAN, effectively transfers knowledge to domains with few target images, outperforming existing methods. In addition, MineGAN can successfully transfer knowledge from multiple pretrained GANs. MineGAN. [ABSTRACT FROM AUTHOR]

• We show that saliency maps can be obtained as a byproduct of image classification. • Our method does not require any saliency ground truth. • We include an extensive study of the effect of center bias on the results. • We get competitive results even when comparing to methods that do use ground truth. Saliency is the perceptual capacity of our visual system to focus our attention (i.e. gaze) on relevant objects instead of the background. So far, computational methods for saliency estimation required the explicit generation of a saliency map, process which is usually achieved via eyetracking experiments on still images. This is a tedious process that needs to be repeated for each new dataset. In the current paper, we demonstrate that is possible to automatically generate saliency maps without ground-truth. In our approach, saliency maps are learned as a side effect of object recognition. Extensive experiments carried out on both real and synthetic datasets demonstrated that our approach is able to generate accurate saliency maps, achieving competitive results when compared with supervised methods. [ABSTRACT FROM AUTHOR]

Self-efficacy for writing is an important motivational factor and considered to predict writing performance. Self-efficacy for narrative writing has been sparsely studied, and few studies focus on the effects of writing intervention on self-efficacy. Additionally, there is a lack of validated measures of self-efficacy for elementary school students. In a previous study, we found that a trained panel rated personal narrative text quality higher for girls than for boys, which led to our aim: to investigate boys' and girls' self-efficacy for narrative writing before and after an intervention, and to explore associations between self-efficacy and text quality. An 18-item self-efficacy scale was developed. Fifty-five fifth-grade students (M 11:2 years, SD 3.7 months) filled out the scale before and after a five-lesson observational learning intervention. Self-efficacy was then related to writing performance as measured by holistic text quality ratings. The students demonstrated strong self-efficacy, which increased significantly post-intervention. Girls and boys demonstrated similar self-efficacy, despite girls' higher text quality. There were moderate correlations between self-efficacy and writing performance pre- and post-intervention. The results support previous findings of strong self-efficacy at this age. The interaction between writing self-efficacy and performance is complex. Young students may not be able to differentiate between self-efficacy, general writing skills, task performance, and self-regulation. Self-efficacy scales should thus be carefully constructed with respect to age, genre, instruction, and to students' general educational context. [ABSTRACT FROM AUTHOR]

This study is concerned with the integration of negation in relation to two models of the processing of negation: (i) the two-step model (Lüdtke et al., 2008), according to which negation involves two representations where negation is ignored in the first representation, and (ii) the pragmatic view (Nieuwland & Kuperberg, 2008), which posits that negation can be integrated without delay if it is used in a natural context. The processing of two negated forms (not authorised and unauthorised) and an affirmative form (authorised) was studied in complex congruent and incongruent contexts. Incongruities in affirmative sentences elicited a biphasic N400–P600. In both types of negated sentences, ERP patterns associated with higher processing difficulties (anterior and central negativities) were observed. The results did not support one or the other model, suggesting that the processing of negation cannot be fully captured by either of them. [ABSTRACT FROM AUTHOR]

This paper addresses the problem of inferring unseen cross-modal image-to-image translations between multiple modalities. We assume that only some of the pairwise translations have been seen (i.e. trained) and infer the remaining unseen translations (where training pairs are not available). We propose mix and match networks, an approach where multiple encoders and decoders are aligned in such a way that the desired translation can be obtained by simply cascading the source encoder and the target decoder, even when they have not interacted during the training stage (i.e. unseen). The main challenge lies in the alignment of the latent representations at the bottlenecks of encoder–decoder pairs. We propose an architecture with several tools to encourage alignment, including autoencoders and robust side information and latent consistency losses. We show the benefits of our approach in terms of effectiveness and scalability compared with other pairwise image-to-image translation approaches. We also propose zero-pair cross-modal image translation, a challenging setting where the objective is inferring semantic segmentation from depth (and vice-versa) without explicit segmentation-depth pairs, and only from two (disjoint) segmentation-RGB and depth-RGB training sets. We observe that a certain part of the shared information between unseen modalities might not be reachable, so we further propose a variant that leverages pseudo-pairs which allows us to exploit this shared information between the unseen modalities. [ABSTRACT FROM AUTHOR]

• We investigated the role of saliency on improving the classification accuracy when the training data is scarce. • We considered adding a saliency branch to an existing CNN architecture (AlexNet, ResNet-50 and ResNet-152). • We validated our approach on the fine-grained object recognition problem. • Experimental results confirmed that our approach is useful for the case when the available training data is scarce. • Our experiments show that there exists a clear correlation (Pearson coefficient) between the performance of saliency methods on standard saliency benchmarks and the performance gain that is obtained when incorporating them in a object recognition pipeline. This paper investigates the role of saliency to improve the classification accuracy of a Convolutional Neural Network (CNN) for the case when scarce training data is available. Our approach consists in adding a saliency branch to an existing CNN architecture which is used to modulate the standard bottom-up visual features from the original image input, acting as an attentional mechanism that guides the feature extraction process. The main aim of the proposed approach is to enable the effective training of a fine-grained recognition model with limited training samples and to improve the performance on the task, thereby alleviating the need to annotate a large dataset. The vast majority of saliency methods are evaluated on their ability to generate saliency maps, and not on their functionality in a complete vision pipeline. Our proposed pipeline allows to evaluate saliency methods for the high-level task of object recognition. We perform extensive experiments on various fine-grained datasets (Flowers, Birds, Cars, and Dogs) under different conditions and show that saliency can considerably improve the network's performance, especially for the case of scarce training data. Furthermore, our experiments show that saliency methods that obtain improved saliency maps (as measured by traditional saliency benchmarks) also translate to saliency methods that yield improved performance gains when applied in an object recognition pipeline. [ABSTRACT FROM AUTHOR]

A match-action cut in feature films connects two shots of a single continuous movement. This type of editing often goes unnoticed and is arguably the most effective form of continuity editing. However, the literature offers little agreement on editing best practice and, by implication, on how our perceptual system deals with disjointed moving images. Studies have suggested that frames should overlap across the cut for the viewer to experience continuity, but also that leaving out frames is preferable, and even that viewers are unable to discriminate such detail. We conducted an experiment to investigate viewer preferences for match-action cuts, using type of cut as well as velocity of movement as predictors and number of overlapping/elliptical frames as the outcome variable. Thirty-nine participants determined the smoothest cut in eight film-clips in a within-subjects design. Surprisingly, we found that average viewer preferences were less than a single frame from a straight cut for all cut-types. We also found that velocity had a small but statistically significant effect on editing preferences. The preference for straight cuts found in the present study runs counter to the idea that perceived continuity across match-action cuts requires objective dis-continuity and suggests that the straight cut provides a simple rule of thumb for film editors. In addition, we interpret the conflicting results from previous studies together with our own findings based on a discrimination task of finding the optimal cut as indicating that human visual perception allows for a window of acceptable continuity cuts centered around the straight cut. [ABSTRACT FROM AUTHOR]

Designing discriminative powerful texture features robust to realistic imaging conditions is a challenging computer vision problem with many applications, including material recognition and analysis of satellite or aerial imagery. In the past, most texture description approaches were based on dense orderless statistical distribution of local features. However, most recent approaches to texture recognition and remote sensing scene classification are based on Convolutional Neural Networks (CNNs). The de facto practice when learning these CNN models is to use RGB patches as input with training performed on large amounts of labeled data (ImageNet). In this paper, we show that Local Binary Patterns (LBP) encoded CNN models, codenamed TEX-Nets, trained using mapped coded images with explicit LBP based texture information provide complementary information to the standard RGB deep models. Additionally, two deep architectures, namely early and late fusion, are investigated to combine the texture and color information. To the best of our knowledge, we are the first to investigate Binary Patterns encoded CNNs and different deep network fusion architectures for texture recognition and remote sensing scene classification. We perform comprehensive experiments on four texture recognition datasets and four remote sensing scene classification benchmarks: UC-Merced with 21 scene categories, WHU-RS19 with 19 scene classes, RSSCN7 with 7 categories and the recently introduced large scale aerial image dataset (AID) with 30 aerial scene types. We demonstrate that TEX-Nets provide complementary information to standard RGB deep model of the same network architecture. Our late fusion TEX-Net architecture always improves the overall performance compared to the standard RGB network on both recognition problems. Furthermore, our final combination leads to consistent improvement over the state-of-the-art for remote sensing scene classification. [ABSTRACT FROM AUTHOR]

Color description is one of the fundamental problems of image understanding. One of the popular ways to represent colors is by means of color names. Most existing work on color names focuses on only the eleven basic color terms of the English language. This could be limiting the discriminative power of these representations, and representations based on more color names are expected to perform better. However, there exists no clear strategy to choose additional color names. We collect a dataset of 28 additional color names. To ensure that the resulting color representation has high discriminative power we propose a method to order the additional color names according to their complementary nature with the basic color names. This allows us to compute color name representations with high discriminative power of arbitrary length. In the experiments we show that these new color name descriptors outperform the existing color name descriptor on the task of visual tracking, person re-identification and image classification. [ABSTRACT FROM AUTHOR]

Aims and objectives: In this study, we investigated crosslinguistic influence in the phonetic systems of simultaneous bilinguals (2L1s) during adulthood. Methodology: Specifically, we analyzed the voice onset time (VOT) of the voiceless stop /k/ in the spontaneous speech of 14 German–French bilinguals who grew up in France or Germany. We looked at both languages, first comparing the groups, second comparing their VOT to their global accent. Data and analysis: The material consisted of interviews, lasting for about half an hour. Findings/conclusions: Most 2L1s showed distinct VOT-ranges in their two languages, even if they were perceived to have a foreign accent in the minority language of their childhood environment. We conclude that the phonetic systems of 2L1s remain separate and stable throughout the lifespan. However, the 2L1s from France had significantly shorter VOTs in German than the 2L1s from Germany, and their speech was overall more accented. These findings are discussed with respect to the role of intra- and extra-linguistic factors. Originality: Our study adds a new perspective to existing VOT studies of bilinguals by using naturalistic speech data and by comparing two groups of 2L1s who have the same language combination but grew up in different countries, which allows us to evaluate the impact of their childhood environment on VOT development. Significance/implications: Language exposure during childhood seems to be beneficial for pronunciation during adulthood. [ABSTRACT FROM AUTHOR]

This study has two goals: First, to give an account of the semantic organization of individually used antonymic adjectives in discourse and second, based on those findings and previous work on antonymic meanings, to contribute to a comprehensive theoretical account of their representation within the framework of Cognitive Linguistics. The hypothesis is that the members of the pairs are used in the same contexts and in the same type of constructions, not only when they co-occur and are used to express binary opposition as shown in previous studies, but also when they do not. The manually coded corpus data from the BNC are analyzed along four semantic parameters: (i) the configuration of the adjectives in terms of gradability, (ii) the way they modify the nominal meanings, i.e., attributively or predicatively (iii) the meaning type of the modified nouns, and (iv) the status of the constructions with respect to whether their meanings are what we refer to as 'basic', metaphorical or metonymical. Correspondence analysis technique is used to identify similarities and differences on the basis of the totality of the data. As predicted, our findings confirm a high degree of pairwise similarity - but also some differences. On the basis of these results, it can be argued that the long-standing controversy within Structuralism between proponents of the co-occurrence hypothesis and the substitutability hypothesis in antonym research is a non-issue. [ABSTRACT FROM AUTHOR]

Describing textures is a challenging problem in computer vision and pattern recognition. The classification problem involves assigning a category label to the texture class it belongs to. Several factors such as variations in scale, illumination and viewpoint make the problem of texture description extremely challenging. A variety of histogram based texture representations exists in literature. However, combining multiple texture descriptors and assessing their complementarity is still an open research problem. In this paper, we first show that combining multiple local texture descriptors significantly improves the recognition performance compared to using a single best method alone. This gain in performance is achieved at the cost of high-dimensional final image representation. To counter this problem, we propose to use an information-theoretic compression technique to obtain a compact texture description without any significant loss in accuracy. In addition, we perform a comprehensive evaluation of pure color descriptors, popular in object recognition, for the problem of texture classification. Experiments are performed on four challenging texture datasets namely, KTH-TIPS-2a, KTH-TIPS-2b, FMD and Texture-10. The experiments clearly demonstrate that our proposed compact multi-texture approach outperforms the single best texture method alone. In all cases, discriminative color names outperforms other color features for texture classification. Finally, we show that combining discriminative color names with compact texture representation outperforms state-of-the-art methods by 7.8%, 4.3% and 5.0% on KTH-TIPS-2a, KTH-TIPS-2b and Texture-10 datasets respectively. [ABSTRACT FROM AUTHOR]

Purpose: To investigate gaze behavior during communication between children with hearing impairment (HI) and normal-hearing (NH) peers. Method: Ten HI–NH and 10 NH–NH dyads performed a referential communication task requiring description of faces. During task performance, eye movements and speech were tracked. Using verbal event (questions, statements, back channeling, and silence) as the predictor variable, group characteristics in gaze behavior were expressed with Kaplan-Meier survival functions (estimating time to gaze-to-partner) and odds ratios (comparing number of verbal events with and without gaze-to-partner). Analyses compared the listeners in each dyad (HI: n = 10, mean age = 12;6 years, mean better ear pure-tone average = 33.0 dB HL; NH: n = 10, mean age = 13;7 years). Results: Log-rank tests revealed significant group differences in survival distributions for all verbal events, reflecting a higher probability of gaze to the partner’s face for participants with HI. Expressed as odds ratios (OR), participants with HI displayed greater odds for gaze-to-partner (ORs ranging between 1.2 and 2.1) during all verbal events. Conclusions: The results show an increased probability for listeners with HI to gaze at the speaker’s face in association with verbal events. Several explanations for the finding are possible, and implications for further research are discussed. [ABSTRACT FROM AUTHOR]

Purpose: To investigate gaze behavior during communication between children with hearing impairment (HI) and normal-hearing (NH) peers. Method: Ten HI–NH and 10 NH–NH dyads performed a referential communication task requiring description of faces. During task performance, eye movements and speech were tracked. Using verbal event (questions, statements, back channeling, and silence) as the predictor variable, group characteristics in gaze behavior were expressed with Kaplan-Meier survival functions (estimating time to gaze-to-partner) and odds ratios (comparing number of verbal events with and without gaze-to-partner). Analyses compared the listeners in each dyad (HI: n = 10, mean age = 12;6 years, mean better ear pure-tone average = 33.0 dB HL; NH: n = 10, mean age = 13;7 years). Results: Log-rank tests revealed significant group differences in survival distributions for all verbal events, reflecting a higher probability of gaze to the partner’s face for participants with HI. Expressed as odds ratios (OR), participants with HI displayed greater odds for gaze-to-partner (ORs ranging between 1.2 and 2.1) during all verbal events. Conclusions: The results show an increased probability for listeners with HI to gaze at the speaker’s face in association with verbal events. Several explanations for the finding are possible, and implications for further research are discussed. [ABSTRACT FROM AUTHOR]

Most existing color constancy algorithms assume uniform illumination. However, in real-world scenes, this is not often the case. Thus, we propose a novel framework for estimating the colors of multiple illuminants and their spatial distribution in the scene. We formulate this problem as an energy minimization task within a conditional random field over a set of local illuminant estimates. In order to quantitatively evaluate the proposed method, we created a novel data set of two-dominant-illuminant images comprised of laboratory, indoor, and outdoor scenes. Unlike prior work, our database includes accurate pixel-wise ground truth illuminant information. The performance of our method is evaluated on multiple data sets. Experimental results show that our framework clearly outperforms single illuminant estimators as well as a recently proposed multi-illuminant estimation approach. [ABSTRACT FROM PUBLISHER]

Abstract: Spatial pyramids have been successfully applied to incorporating spatial information into bag-of-words based image representation. However, a major drawback is that it leads to high dimensional image representations. In this paper, we present a novel framework for obtaining compact pyramid representation. First, we investigate the usage of the divisive information theoretic feature clustering (DITC) algorithm in creating a compact pyramid representation. In many cases this method allows us to reduce the size of a high dimensional pyramid representation up to an order of magnitude with little or no loss in accuracy. Furthermore, comparison to clustering based on agglomerative information bottleneck (AIB) shows that our method obtains superior results at significantly lower computational costs. Moreover, we investigate the optimal combination of multiple features in the context of our compact pyramid representation. Finally, experiments show that the method can obtain state-of-the-art results on several challenging data sets. [Copyright &y& Elsevier]

Computational color constancy is a fundamental prerequisite for many computer vision applications. This paper presents a survey of many recent developments and state-of-the-art methods. Several criteria are proposed that are used to assess the approaches. A taxonomy of existing algorithms is proposed and methods are separated in three groups: static methods, gamut-based methods, and learning-based methods. Further, the experimental setup is discussed including an overview of publicly available datasets. Finally, various freely available methods, of which some are considered to be state of the art, are evaluated on two datasets. [ABSTRACT FROM PUBLISHER]

Abstract: We performed a set of experiments dealing with the interpretation of novel Place–Food noun–noun compounds (NNCs) denoting fictive food products, in Danish. These were used in order to address concerns of linguistic theory, in particular the interrelation between semantics and pragmatics, psycholinguistics—investigating the role of sentential context for the interpretation of noun–noun compounds, and consumer oriented food labeling research. In the first experiment, a significant proportion of the novel NNCs was found to have a pre-contextually biased interpretation, in the majority of the cases conforming to what is usually considered the “default” interpretation for the studied type of compounds: FOOD ORIGINATING IN PLACE. In a subsequent experiment, this was however found to be easily overturned when sentential context supported what was initially found to be the least preferred interpretation. This contrasts with previous research comparing the role of such context on conventional NNCs. This suggests a qualitative difference between novel pre-contextually biased NNCs and conventional NNCs, and has implications for the semantics–pragmatics boundary, which we discuss. Furthermore, the three types of interpretations offered to the participants: ORIGINATING IN, RECIPE FROM and REMINDS OF, were shown to form an implicational hierarchy, which could explain the “asymmetric” preferences attributed to them by the participants in the experiments. [Copyright &y& Elsevier]

This paper examines data from English, Swedish and German in order to find a theoretical distribution that describes the observed relation between word length and frequency. In Swedish and English, most word tokens consist of three letters only, while shorter or longer words occur less frequently. We found that the equation with the general form f exp = a * L b * c L (a variant of the so-called gamma distribution) approximates the observed frequencies reasonably well. This formula incorporates both the fact that the number of possible words increases with word length, and the fact that longer words tend to be avoided, presumably because they are uneconomic. To our knowledge this formula has not been proposed to describe word frequency data. We examined frequency distributions of word length in Swedish and English, and explored different variants of the equation by systematically varying the a, b and c parameters. Subsequently, we also applied the formula to the frequency distribution of sentence length in English, and found an almost perfect fit for a corpus consisting of different text genres. Moreover, the data showed that the formula can be used to distinguish between different kinds of text genres. [ABSTRACT FROM AUTHOR]

This study addresses the postulate of non-conventionality of pantomime, inherent in pantomimic scenarios of language origin. Since lack of semiotic conventions does not preclude micro-conventions resulting from cultural differences, pantomimes should be easier to interpret when the actor and recipient share the same culture than between two different cultures. In the study, Italian and Polish amateur "actors" re-enacted transitive events from a matrix of cartoon-like drawings. Randomly selected clips were matched by Polish and Italian participants to the corresponding drawings. We found no difference in the number of correct guesses when the actors and matchers were from the same versus from different cultures. We discuss this result in the context of the core assumptions of pantomimic scenarios of language origin. • Pantomimic scenarios of language origins assume that pantomime is robustly iconic and free from communicative conventions. • Lack of conventions does not preclude micro-conventions, thus pantomime should be more successful within cultures. • This was tested in a study with Italian and Polish participants. • Communicative success of pantomimes did not differ between same-culture and different-culture conditions. • Italian pantomimes were better guessed by both Italian and Polish participants. [ABSTRACT FROM AUTHOR]

We derive the decomposition of the anisotropic Gaussian in a one-dimensional (1-D) Gauss filter in the x-direction followed by a 1-D filter in a non-orthogonal direction ψ So also the anisotropic Gaussian can be decomposed by dimension. This appears to be extremely efficient from a computing perspective. An implementation scheme for normal convolution and for recursive filtering is proposed. Also directed derivative filters are demonstrated. For the recursive implementation, filtering an 512 × 512 image is performed within 40 msec on a current state of the art PC, gaining over 3 times in performance for a typical filter, independent of the standard deviations and orientation of the filter. Accuracy of the filters is still reasonable when compared to truncation error or recursive approximation error. The anisotropic Gaussian filtering method allows fast calculation of edge and ridge maps, with high spatial and angular accuracy. For tracking applications, the normal anisotropic convolution scheme is more advantageous, with applications in the detection of dashed lines in engineering drawings. The recursive implementation is more attractive in feature detection applications, for instance in affine invariant edge and ridge detection in computer vision. The proposed computational filtering method enables the practical applicability of orientation scale-space analysis. [ABSTRACT FROM AUTHOR]

Modern computer vision requires processing large amounts of data, both while training the model and/or during inference, once the model is deployed. Scenarios where images are captured and processed in physically separated locations are increasingly common (e.g. autonomous vehicles, cloud computing, smartphones). In addition, many devices suffer from limited resources to store or transmit data (e.g. storage space, channel capacity). In these scenarios, lossy image compression plays a crucial role to effectively increase the number of images collected under such constraints. However, lossy compression entails some undesired degradation of the data that may harm the performance of the downstream analysis task at hand, since important semantic information may be lost in the process. Moreover, we may only have compressed images at training time but are able to use original images at inference time (i.e. test), or vice versa, and in such a case, the downstream model suffers from covariate shift. In this paper, we analyze this phenomenon, with a special focus on vision-based perception for autonomous driving as a paradigmatic scenario. We see that loss of semantic information and covariate shift do indeed exist, resulting in a drop in performance that depends on the compression rate. In order to address the problem, we propose dataset restoration, based on image restoration with generative adversarial networks (GANs). Our method is agnostic to both the particular image compression method and the downstream task; and has the advantage of not adding additional cost to the deployed models, which is particularly important in resource-limited devices. The presented experiments focus on semantic segmentation as a challenging use case, cover a broad range of compression rates and diverse datasets, and show how our method is able to significantly alleviate the negative effects of compression on the downstream visual task. [ABSTRACT FROM AUTHOR]

Observational learning is a successful method for improving writing skills in various genres. We explore effects of a five lesson intervention series based on peer observation. Fifty-five Swedish 5th-grade students aged 10–12 years followed this intervention programme. The students watched short film clips with peers working with texts. Each lesson was organised according to a theme: reader's perception of the text, ordering of events, how to begin a story, how to end a story and how to edit a text. The students wrote four texts during the intervention. The quality of these texts was assessed by a panel of trained raters. Additionally, the language and reading comprehension and working memory capacity were tested. The results show that average text quality had significantly improved at the end of the intervention, and that this improvement was modulated by reading and language comprehension. Three months later, however, text quality was significantly decreased. [ABSTRACT FROM AUTHOR]

• We discuss whether Chinese given names can be identified as female or male. • The role of specific written Chinese characters in gender identification is discussed. • Gender identification is subject to vocalic sound symbolism. • Reduplication plays a role in gender identification. • We test native speakers' ability to guess gender in different conditions. In this paper we discuss a number of factors that bear on the question if a Chinese given name is more likely to refer to a female or a male. In some cases this can be determined (with some degree of confidence) – in others it cannot. We identify the relevant factors as 1) gender-identifying characters or radicals; 2) sound symbolism and 3) reduplication. We consider the relations between these factors, and test our predictions in a psycholinguistic experiment with native speakers, for both written and spoken Chinese. [ABSTRACT FROM AUTHOR]

Human communication can be either monosemiotic or polysemiotic, depending on whether it combines ensembles of representations from one or more semiotic systems such as language, gesture and depiction. Each semiotic system has its unique storytelling potentials, which makes intersemiotic translation from one system to another challenging. We investigated the influence of the source semiotic system, realised in speech and a sequence of pictures, respectively, on the way the same story was retold using speech and co-speech gestures. The story was the content of the picture book Frog, Where Are You?. A group of Finnish speakers saw the story in pictures, and another group heard it in matched oral narration. Each participant retold the story to an addressee and all narrations were video-recorded and analysed for both speech and gestures. Given the high degree of iconicity in depiction, we expected more iconic gestures (especially enactments) in the narratives translated from pictures than in those translated from speech. Conversely, we expected greater narrative coherence in the narratives translated from speech. The results showed that more iconic gestures were produced in the narratives translated from speech, but these were primarily not from the enactment subtype. As expected, iconic enactments were more frequent in the narratives translated from the story presented in pictures. The narratives produced by participants who had only heard the story did not have a greater variety of connective devices, yet the type of devices differed slightly between the groups. Together with some additional differences between the groups that had not been anticipated, the results indicate that a story presented in different semiotic systems tends to be translated into different polysemiotic narratives. [ABSTRACT FROM AUTHOR]

Leonard Talmy's influential binary motion event typology has encountered four main challenges: (a) additional language types; (b) extensive "type-internal" variation; (c) the role of other relevant form classes than verbs and "satellites;" and (d) alternative definitions of key semantic concepts like Motion, Path and Manner. After reviewing these issues, we show that the theory of Holistic Spatial Semantics provides analytical tools for their resolution. In support, we present an analysis of motion event descriptions by speakers of two languages that are troublesome for the original typology: Thai (Tai-Kadai) and Telugu (Dravidian), based on the Frog-story elicitation procedure. Despite some apparently similar typological features, the motion event descriptions in the two languages were found to be significantly different. The Telugu participants used very few verbs in contrast to extensive case marking to express Path and nominals to express Region and Landmark, while the Thai speakers relied largely on serial verbs for expressing Path and on prepositions for expressing Region. Combined with previous research in the field, our findings imply (at least) four different clusters of languages in motion event typology with Telugu and Thai as representative of two such clusters, languages like French and Spanish representing a third cluster, and Swedish and English a fourth. This also implies that many other languages like Italian, Bulgarian, and Basque will appear as "mixed languages," positioned between two or three of these clusters. [ABSTRACT FROM AUTHOR]

In this eye-tracking and drawing study, we investigate the perceptual grounding of different types of spatial dimensions such as dense – sparse and top – bottom , focusing both on the participants' experiences of the opposite regions, e.g., O1: dense ; O2: sparse , and the region that is experienced as intermediate, e.g., INT: neither dense nor sparse . Six spatial dimensions expected to have three different perceptual structures in terms of the point and range nature of O1, INT and O2 were analysed. Presented with images, the participants were instructed to identify each region (O1, INT, O2), first by looking at the region, and then circumscribing it using the computer mouse. We measured the eye movements, identification times and various characteristics of the drawings such as the relative size of the three regions, overlaps and gaps. Three main results emerged. Firstly, generally speaking, intermediate regions were not different from the poles on any of the indicators: overall identification times, number of fixations, and locations. Some differences emerged with regard to the duration of fixations for point INTs and the number of fixations for range INTs between two range poles (O1, O2). Secondly, the analyses of the fixation locations showed that the poles support the identification of the intermediate region as much as the intermediate region supports the identification of the poles. Finally, the relative size of the three areas selected in the drawing task were consistent with the classification of the regions as points or ranges. The analyses of the gaps and the overlaps between the three areas showed that the intermediate is neither O1 nor O2, but an entity in its own right. [ABSTRACT FROM AUTHOR]

According to film mythology, the Soviet filmmaker Lev Kuleshov conducted an experiment in which he combined a close-up of an actor’s neutral face with three different emotional contexts: happiness, sadness, and hunger. The viewers of the three film sequences reportedly perceived the actor’s face as expressing an emotion congruent with the given context. It is not clear, however, whether or not the so-called “Kuleshov effect” really exists. The original film footage is lost and recent attempts at replication have produced either conflicting or unreliable results. The current paper describes an attempt to replicate Kuleshov’s original experiment using an improved experimental design. In a behavioral and eye tracking study, 36 participants were each presented with 24 film sequences of neutral faces across six emotional conditions. For each film sequence, the participants were asked to evaluate the emotion of the target person in terms of valence, arousal, and category. The participants’ eye movements were recorded throughout. The results suggest that some sort of Kuleshov effect does in fact exist. For each emotional condition, the participants tended to choose the appropriate category more frequently than the alternative options, while the answers to the valence and arousal questions also went in the expected directions. The eye tracking data showed how the participants attended to different regions of the target person’s face (in light of the intermediate context), but did not reveal the expected differences between the emotional conditions. [ABSTRACT FROM AUTHOR]