Showing papers by "School for Advanced Studies in the Social Sciences published in 2018"

Linguistic Unit Discovery from Multi-Modal Inputs in Unwritten Languages: Summary of the “Speaking Rosetta” JSALT 2017 Workshop

Abstract: We summarize the accomplishments of a multi-disciplinary workshop exploring the computational and scientific issues surrounding the discovery of linguistic units (subwords and words) in a language without orthography. We study the replacement of orthographic transcriptions by images and/or translated text in a well-resourced language to help unsupervised discovery from raw speech.

Cerebellar learning using perturbations.

Abstract: The cerebellum aids the learning of fast, coordinated movements. According to current consensus, erroneously active parallel fibre synapses are depressed by complex spikes signalling movement errors. However, this theory cannot solve the credit assignment problem of processing a global movement evaluation into multiple cell-specific error signals. We identify a possible implementation of an algorithm solving this problem, whereby spontaneous complex spikes perturb ongoing movements, create eligibility traces and signal error changes guiding plasticity. Error changes are extracted by adaptively cancelling the average error. This framework, stochastic gradient descent with estimated global errors (SGDEGE), predicts synaptic plasticity rules that apparently contradict the current consensus but were supported by plasticity experiments in slices from mice under conditions designed to be physiological, highlighting the sensitivity of plasticity studies to experimental conditions. We analyse the algorithm's convergence and capacity. Finally, we suggest SGDEGE may also operate in the basal ganglia.

Virtual navigation tested on a mobile app is predictive of real-world wayfinding navigation performance

Abstract: Virtual reality environments presented on tablets and smartphones have potential to aid the early diagnosis of conditions such as Alzheimer9s dementia by quantifying impairments in navigation performance. However, it is unclear whether performance on mobile devices can predict navigation errors in the real world. We compared the performance of 60 participants (30 females, 18-35 years old) at wayfinding and path integration tasks designed in our mobile app `Sea Hero Quest9 with their performance at similar tasks in a real-world environment. We first performed this experiment in the streets of London (UK) and replicated it in Paris (France). In both cities, we found a significant correlation between virtual and real-world wayfinding performance and a male advantage in both environments, although smaller in the real world (Cohen9s d in the game = 0.89, in the real world = 0.59). Results in London and Paris were highly similar, and controlling for familiarity with video games did not change the results. The strength of the correlation between real world and virtual environment increased with the difficulty of the virtual wayfinding task, indicating that Sea Hero Quest does not merely capture video gaming skills. The fact that the Sea Hero Quest wayfinding task has real-world ecological validity constitutes a step toward controllable, sensitive, safe, low-cost, and easy to administer digital cognitive assessment of navigation ability.

Symbouki: a meta-analysis on the emergence of sound symbolism in early language acquisition.

Abstract: Adults and toddlers systematically associate pseudowords such as "bouba" and "kiki" with round and spiky shapes, respectively, a sound symbolic phenomenon known as the "bouba-kiki effect". To date, whether this sound symbolic effect is a property of the infant brain present at birth or is a learned aspect of language perception remains unknown. Yet, solving this question is fundamental for our understanding of early language acquisition. Indeed, an early sensitivity to such sound symbolic associations could provide a powerful mechanism for language learning, playing a bootstrapping role in the establishment of novel sound-meaning associations. The aim of the present meta-analysis (SymBouKi) is to provide a quantitative overview of the emergence of the bouba-kiki effect in infancy and early childhood. It allows a high-powered assessment of the true sound symbolic effect size by pooling over the entire set of 11 extant studies (six published, five unpublished), entailing data from 425 participants between 4 and 38 months of age. The quantitative data provide statistical support for a moderate, but significant, sound symbolic effect. Further analysis found a greater sensitivity to sound symbolism for bouba-type pseudowords (i.e., round sound-shape correspondences) than for kiki-type pseudowords (i.e., spiky sound-shape correspondences). For the kiki-type pseudowords, the effect emerged with age. Such discrepancy challenges the view that sensitivity to sound symbolism is an innate language mechanism rooted in an exuberant interconnected brain. We propose alternative hypotheses where both innate and learned mechanisms are at play in the emergence of sensitivity to sound symbolic relationships.

XNMT: The eXtensible Neural Machine Translation Toolkit

Abstract: This paper describes XNMT, the eXtensible Neural Machine Translation toolkit. XNMT distin- guishes itself from other open-source NMT toolkits by its focus on modular code design, with the purpose of enabling fast iteration in research and replicable, reliable results. In this paper we describe the design of XNMT and its experiment configuration system, and demonstrate its utility on the tasks of machine translation, speech recognition, and multi-tasked machine translation/parsing. XNMT is available open-source at this https URL

A Neuromathematical Model for Geometrical Optical Illusions

Abstract: Geometrical optical illusions have been object of many studies due to the possibility they offer to understand the behavior of low-level visual processing. They consist in situations in which the perceived geometrical properties of an object differ from those of the object in the visual stimulus. Starting from the geometrical model introduced by Citti and Sarti (J Math Imaging Vis 24(3):307---326, 2006), we provide a mathematical model and a computational algorithm which allows to interpret these phenomena and to qualitatively reproduce the perceived misperception.

The semantic roots of positive polarity: epistemic modal verbs and adverbs in English, Greek and Italian

Abstract: Epistemic modal verbs and adverbs of necessity are claimed to be positive polarity items. We study their behavior by examining modal spread, a phenomenon that appears redundant or even anomalous, since it involves two apparent modal operators being interpreted as a single modality. We propose an analysis in which the modal adverb is an argument of the MUST modal, providing a meta-evaluation $$\mathcal {O}$$ which ranks the Ideal, stereotypical worlds in the modal base as better possibilities than the Non-Ideal worlds in it. MUST and possibility modals differ in that the latter have an empty $$\mathcal {O}$$ , a default that can be negotiated. Languages vary in the malleability of this parameter. Positive polarity is derived as a conflict between the ranking imposed by $$\mathcal {O}$$ —which requires that the Ideal worlds be better possibilities than Non-Ideal worlds—and the effect of higher negation which renders the Ideal set non-homogenous. Applying the ordering over such a non-homogeneous set would express preference towards both p and $$\lnot p $$ worlds thus rendering the sentence uninformative. Negative polarity MUST and possibility modals, on the other hand, contain an empty $$\mathcal {O}$$ , application of higher negation therefore poses no problem. This account is the first to connect modal spread to positive polarity of necessity modals, and captures the properties of both in a unified analysis.

Factors associated with the introduction of complementary feeding in the French ELFE cohort study

Abstract: The objectives of this study were to estimate the age of complementary feeding introduction (CFI) and investigate the related health, demographic, and socio-economic factors. Analyses were based on 10,931 infants from the French national birth cohort ELFE, born in 2011. Health, demographic, and socio-economic data concerning infants and parents were collected at birth (face-to-face interviews and medical records) and 2 months (telephone interviews). Data on milk feeding and CFI practices were collected at birth and 2 months then monthly from 3 to 10 months using online or paper questionnaires. The associations between both health and social factors and CFI age were tested by multivariable multinomial logistic regressions. The mean CFI age was 5.2 ± 1.2 months; 26% of the infants started complementary feeding before 4 months of age (CF 6 months). CF 6 months. Couples in which fathers were born in France and mothers were not born in France were less likely to introduce CF > 6 months. CF < 4 months occurred in more than 25% of the cases. It is important to continue promoting clear CFI recommendations, especially in smoking, overweight, young, not born in France, and nonbreastfeeding mothers.

Moments in Collaboration: Experiments in Concept Work

Abstract: There is an increasing focus among anthropologists on the theme of collaboration with the people they work with and with other disciplines in the university space. Frequently justified in political terms of participation, there is often less attention paid to the conceptual work in and of collaboration. In opposition to the attention given to the processes of exchange during fieldwork, there is rarely a description of the actual forms and practices created for such collective conceptual work and thinking-processes in extra-fieldwork situations. In this article, we report on an experiment in collaborative concept work at Berkeley known as ‘the Labinar'. We address a lacuna in the literature on collaboration by providing a description of how collective conceptual work can be given form and sustained with specific practices. We argue for understanding concepts as not only discursive but also as non-discursive entities, created through and emerging as objects and practices of inquiry. The article focu...

The role of governance networks in building metropolitan scale

Abstract: The role of governance networks in building metropolitan scale – Territory, Politics, Governance. The broad aim of this comparative study is to examine the relationship between governance networks and the emergence (or lack thereof) of metropolitan scales through an analysis of metropolitan development-policy processes. It explores the characteristics and substance of policies that purport to be metropolitan in scope through a set of six case studies of global city-regions lacking a formal metropolitan-scale government: Berlin, Delhi, New York, Paris, Rome and Shenzhen. This is done to obtain a better sense of the networks, strategies and approaches used in various contexts to tackle boundary-spanning issues in regions. Three paired case studies analyse what interests and actors were involved, how central each actor was to the policy process, and what territorial scales and interests dominated to identify commonalities across cases and to look for evidence of the emergence of new actors in metropo...

Distributed Domination on Graph Classes of Bounded Expansion

Abstract: oindent We provide a new constant factor approximation algorithm for the (connected) \mboxdistance- r dominating set problem on graph classes of bounded expansion. Classes of bounded expansion include many familiar classes of sparse graphs such as planar graphs and graphs with excluded (topological) minors, and notably, these classes form the most general subgraph closed classes of graphs for which a sequential constant factor approximation algorithm for the distance- r dominating set problem is currently known. Our algorithm can be implemented in the \congestbc model of distributed computing and uses $Oof(r^2 log n)$ communication rounds. % Our techniques, which may be of independent interest, are based on a distributed computation of sparse neighborhood covers of small radius on bounded expansion classes. We show how to compute an r -neighborhood cover of radius~$2r$ and overlap $f(r)$ on every class of bounded expansion in $Oof(r^2log n)$ communication rounds for some function~ f .% in the \congestbc model. % Finally, we show how to use the greater power of the local model to turn any distance- r dominating set into a constantly larger connected distance- r dominating set in $3r+1$ rounds on any class of bounded expansion. Combining this algorithm, e.g., with the constant factor approximation algorithm for dominating sets on planar graphs of Lenzen et al.\ gives a constant factor approximation algorithm for connected dominating sets on planar graphs in a constant number of rounds in the local model, where the approximation ratio is only $6$ times larger than that of Lenzen et al.'s algorithm.

Where am I in virtual reality

Abstract: It is currently not well understood whether people experience themselves to be located in one or more specific part(s) of their body. Virtual reality (VR) is increasingly used as a tool to study aspects of bodily perception and self-consciousness, due to its strong experimental control and ease in manipulating multi-sensory aspects of bodily experience. To investigate where people self-locate in their body within virtual reality, we asked participants to point directly at themselves with a virtual pointer, in a VR headset. In previous work employing a physical pointer, participants mainly located themselves in the upper face and upper torso. In this study, using a VR headset, participants mainly located themselves in the upper face. In an additional body template task where participants pointed at themselves on a picture of a simple body outline, participants pointed most often to the upper torso, followed by the (upper) face. These results raise the question as to whether head-mounted virtual reality might alter where people locate themselves making them more “head-centred”.

The ACLEW DiViMe: An Easy-to-use Diarization Tool.

Abstract: We present “DiViMe”, an open-source virtual machine aimed at packaging speech technology for real-life data, and developed in the context of the “Analyzing Children’s Language Environments across the World” Project. This first release focuses on Speech Activity Detection, Speaker Diarization, and their evaluation. The present paper introduces the set of included tools and the current workflow, which is focused on making minimal assumptions regarding users’ technical skills. Additionally, we show how the current DiViMe tools fare against three sets of challenging data. In a first experiment, we look at performance with samples extracted from daylong recordings gathered using the LENA system from English-learning children. We find that the performance of the tools currently in DiViMe is not far from that achieved by the LENA proprietary software. In a second experiment, we generalize to other samples of child-centered daylong files, gathered with non-LENA hardware from non-English-learning children, showing that performance does not degrade in this condition. Finally, we report on performance in the DiHARD 2018 Challenge Test Data. Originally conceived in the “Speech Recognition Virtual Kitchen”, DiViMe is a promising platform for packaging speech technology tools for widespread re-use, with potential impact on both fundamental and applied speech and language research.

Sampling Strategies in Siamese Networks for Unsupervised Speech Representation Learning.

Abstract: Recent studies have investigated siamese network architectures for learning invariant speech representations using same-different side information at the word level. Here we investigate systematically an often ignored component of siamese networks: the sampling procedure (how pairs of same vs. different tokens are selected). We show that sampling strategies taking into account Zipf's Law, the distribution of speakers and the proportions of same and different pairs of words significantly impact the performance of the network. In particular, we show that word frequency compression improves learning across a large range of variations in number of training pairs. This effect does not apply to the same extent to the fully unsupervised setting, where the pairs of same-different words are obtained by spoken term discovery. We apply these results to pairs of words discovered using an unsupervised algorithm and show an improvement on state-of-the-art in unsupervised representation learning using siamese networks.

Talker diarization in the wild: The case of child-centered daylong audio-recordings

Abstract: Speaker diarization (answering ’who spoke when’) is a widely researched subject within speech technology. Numerous experiments have been run on datasets built from broadcast news, meeting data, and call centers—the task sometimes appears close to being solved. Much less work has begun to tackle the hardest diarization task of all: spontaneous conversations in real-world settings. Such diarization would be particularly useful for studies of language acquisition, where researchers investigate the speech children produce and hear in their daily lives. In this paper, we study audio gathered with a recorder worn by small children as they went about their normal days. As a result, each child was exposed to different acoustic environments with a multitude of background noises and a varying number of adults and peers. The inconsistency of speech and noise within and across samples poses a challenging task for speaker diarization systems, which we tackled via retraining and data augmentation techniques. We further studied sources of structured variation across raw audio files, including the impact of speaker type distribution, proportion of speech from children, and child age on diarization performance. We discuss the extent to which these findings might generalize to other samples of speech in the wild.

Food rejection and the development of food category-based induction in 2–6 years old children

Abstract: We studied children’s inductive inferences within the domain of food categories. There has so far been little research on inductive reasoning about food among children, despite the theoretical and ...

Shopping and cooking: the organization of food practices, at the crossing of access to food stores and household properties in France

Abstract: This paper explores different ways of organizing food practices in shopping, cooking, and managing leftovers and shows how these relate to sustainability. We conducted an ethnographic study in France based on in-depth and repeated interviews with around 30 ‘ordinary’ consumers aged between 30 and 87 years. We analyzed the interviews using a practice–theory approach and distinguished meanings, materials, and skills linked to food products and eating. We identify four patterns of everyday food practices, each coherently linking specific ways of provisioning, storing, cooking, waste sorting, and other practices. We show how households adopt patterns according to their social characteristics and place of residence and how they switch from one pattern to another according to circumstances. Each pattern comprises some sustainable practices, although not always at the same level. We highlight not only the role of material infrastructure in framing access to food products, but also the necessity to consider temporal organization, financial resources, household size, and social position to understand food practices. Food practices also differ according to definitions of proper eating, which may vary in the long run according to life course events, and in the short run according to the context of meals. We conclude by discussing different ways to promote more sustainable eating.

A Geometric Model of Multi-scale Orientation Preference Maps via Gabor Functions

Abstract: In this paper we present a new model for the generation of orientation preference maps in the primary visual cortex (V1), considering both orientation and scale features. First we undertake to model the functional architecture of V1 by interpreting it as a principal fiber bundle over the 2-dimensional retinal plane by introducing intrinsic variables orientation and scale. The intrinsic variables constitute a fiber on each point of the retinal plane and the set of receptive profiles of simple cells is located on the fiber. Each receptive profile on the fiber is mathematically interpreted as a rotated Gabor function derived from an uncertainty principle. The visual stimulus is lifted in a 4-dimensional space, characterized by coordinate variables, position, orientation and scale, through a linear filtering of the stimulus with Gabor functions. Orientation preference maps are then obtained by mapping the orientation value found from the lifting of a noise stimulus onto the 2-dimensional retinal plane. This corresponds to a Bargmann transform in the reducible representation of the $$\text {SE}(2)=\mathbb {R}^2\times S^1$$ group. A comparison will be provided with a previous model based on the Bargmann transform in the irreducible representation of the $$\text {SE}(2)$$ group, outlining that the new model is more physiologically motivated. Then, we present simulation results related to the construction of the orientation preference map by using Gabor filters with different scales and compare those results to the relevant neurophysiological findings in the literature.

Weak Ties and Expertise: Crossing Technological Boundaries

Abstract: In this article, we question how new technological traits can penetrate cohesive social groups and spread. Based on ethnographic narratives and following studies in sociology, the hypothesis is that not only weak ties are important for linking otherwise unconnected groups and introducing new techniques but also that expertise is required. In order to test this hypothesis, we carried out a set of field experiments in northern India where the kiln has been adopted recently. Our goal was to measure the degree of expertise of the potters distributed between early and late adopters of the kiln. Our results are discussed in the light of oral interviews. Our conclusions suggest that expertise is a necessary, albeit not sufficient, condition for weak ties to act as bridges and thereby for new techniques to spread. As an example, they explain how turntables could have been adopted by potters from the northern Levant during the third millennium BC.

The new watchdogs' vision of science: A roundtable with Ivan Oransky ( Retraction Watch) and Brandon Stell ( PubPeer).

Abstract: On March 3rd, 2016, the authors of this note hosted a conference entitled ‘Destabilized Science’ at the University of California, Los Angeles, to which we invited two representatives of core actors within the new science watchdog pack: Ivan Oransky, co-founder in 2010 of Retraction Watch, and Brandon Stell, co-founder in 2012 of PubPeer. After the formal conference, we organized a roundtable to discuss these invitees’ experience and their vision of contemporary science. Mario Biagioli (University of California, Davis), Michael Chwe (UCLA) and Aaron Panofsky (UCLA) participated to the conversation. An edited transcript of the discussion and a short podcast version are being published on Transmissions (ssstransmissions.org) the new blog associated with Social Studies of Science.

Information Circulation in times of Ebola: Twitter and the Sexual Transmission of Ebola by Survivors

Abstract: INTRODUCTION The 2013-2015 outbreak of Ebola was by far the largest to date, affecting Guinea, Liberia, Sierra Leone, and secondarily, Nigeria, Senegal and the United States. Such an event raises questions about the circulation of health information across social networks. This article presents an analysis of tweets concerning a specific theme: the sexual transmission of the virus by survivors, at a time when there was a great uncertainty about the duration and even the possibility of such transmission. METHODS This article combines quantitative and qualitative analysis. From a sample of 50,000 tweets containing the words "Ebola" in French and English, posted between March 15 and November 8, 2014, we created a graphic representation of the number of tweets over time, and identified two peaks: the first between July 27 and August 16, 2014 (633 tweets) and the second between September 28 and November 8, 2014 (2,577 tweets). This sample was divided into two parts, and every accessible publication was analyzed and coded according to the authors' objectives, feelings expressed and/or publication type. RESULTS While the results confirm the significant role played by mainstream media in disseminating information, media did not create the debate around the sexual transmission of Ebola and Twitter does not fully reflect mainstream media contents. Social media rather work like a "filter": in the case of Ebola, Twitter preceded and amplified the debate with focusing more than the mainstream media on the sexual transmission, as expressed in jokes, questions and criticism. DISCUSSION Online debates can of course feed on journalistic or official information, but they also show great autonomy, tinged with emotions or criticisms. Although numerous studies have shown how this can lead to rumors and disinformation, our research suggests that this relative autonomy makes it possible for Twitter users to bring into the public sphere some types of information that have not been widely addressed. Our results encourage further research to understand how this "filter" works during health crises, with the potential to help public health authorities to adjust official communications accordingly. Without a doubt, the health authorities would be well advised to put in place a special watch on the comments circulating on social media (in addition to that used by the health monitoring agencies).

XNMT: the eXtensible Neural Machine Translation toolkit

Abstract: This paper describes XNMT, the eXtensible Neural Machine Translation toolkit. XNMT distin- guishes itself from other open-source NMT toolkits by its focus on modular code design, with the purpose of enabling fast iteration in research and replicable, reliable results. In this paper we describe the design of XNMT and its experiment configuration system, and demonstrate its utility on the tasks of machine translation, speech recognition, and multi-tasked machine translation/parsing. XNMT is available open-source at this https URL

Living in an Egyptian Oasis: Reconstruction of the Holocene Archaeological Sequence in Kharga

Abstract: Within the Egyptian Western Desert, the Kharga Oasis was a particularly favourable and highly frequented area during the Holocene thanks to a permanent ground water supply. The high concentration and accessibility of archaeological sites here allows the cultural and economic changes to be documented at a local scale, when northeastern Africa was experiencing strong climatic variations and the transition to a food production economy. An expansive program led by the Institut francais d’archeologie orientale (IFAO) which included systematic survey, sampling collections, tests and stratigraphic excavations has dramatically increased the amount of data available concerning the prehistory of the Kharga Oasis. Typo-technological analyses focusing on the lithic industries, and other aspects of material culture, plus a series of 14C dates have led to the definition of four main successive cultural phases: Kharga A, B, C and D. This paper describes this proposal of a new archaeological sequence for human occupation in the Kharga Oasis and discusses its integration in the wider frame of Egypt and the Eastern Sahara.

Revisiting the “Ozone Hole” Metaphor: From Observational Window to Global Environmental Threat

Abstract: In 1985, just over 30 years ago, the “ozone hole” made its appearance in the press as a truly global environmental threat. As one of the most important environmental issues of the twentieth century, the “ozone hole” is also a remarkable metaphorical, visual, and imaginary construction. This essay examines the historical trajectory of the famous “ozone hole” from its birth within the astronomical community at the beginning of the twentieth century to its contemporary framing as a global environmental threat. The article provides evidence why metaphors constitute a valuable object of historically informed studies of scientific practice, and shows in particular how metaphorical landscapes shift over time, mapping at the same time larger social and political developments. The essay ends by showing how scientific images and metaphorical framings interact and how they shape scientific and popular discourse on nature, as well as our understanding of the global environment.