Confidently making progress on multilingual modeling requires challenging, trustworthy evaluations. We present TyDi QA—a question answering dataset covering 11 typologically diverse languages with ...204K question-answer pairs. The languages of TyDi QA are diverse with regard to their typology—the set of linguistic features each language expresses—such that we expect models performing well on this set to generalize across a large number of the world’s languages. We present a quantitative analysis of the data quality and example-level qualitative linguistic analyses of observed language phenomena that would not be found in English-only corpora. To provide a realistic information-seeking task and avoid priming effects, questions are written by people who want to know the answer, but don’t know the answer yet, and the data is collected directly in each language without the use of translation.
Pipelined NLP systems have largely been superseded by end-to-end neural modeling, yet nearly all commonly used models still require an explicit tokenization step. While recent tokenization approaches ...based on data-derived subword lexicons are less brittle than manually engineered tokenizers, these techniques are not equally suited to all languages, and the use of any fixed vocabulary may limit a model’s ability to adapt. In this paper, we present
, a neural encoder that operates directly on character sequences—without explicit tokenization or vocabulary—and a pre-training strategy that operates either directly on characters or optionally uses subwords as a soft inductive bias. To use its finer-grained input effectively and efficiently,
combines downsampling, which reduces the input sequence length, with a deep transformer stack, which encodes context.
outperforms a comparable m
model by 5.7 F1 on
, a challenging multilingual benchmark, despite having fewer model parameters.
One key responsibility of leaders involves crafting and communicating two types of messages—visions and values—that help followers understand the ultimate purpose of their work. Although scholars ...have long considered how leaders communicate visions and values to establish a sense of purpose, they have overlooked how these messages can be used to establish a shared sense of purpose, which is achieved when multiple employees possess the same understanding of the purpose of work. In this research, we move beyond the traditional focus on leader rhetoric and individual cognition to examine leader rhetoric and shared cognition. We suggest that a specific combination of messages—a large amount of vision imagery combined with a small number of values—will boost performance more than other combinations because it triggers a shared sense of the organization's ultimate goal, and, in turn, enhances coordination. We found support for our predictions in an archival study of 151 hospitals and an experiment with 62 groups of full-time employees. In light of these findings, we conducted exploratory analyses and discovered two dysfunctional practices: leaders tend to (1) communicate visions without imagery and (2) over-utilize value-laden rhetoric.
Traditional prognostic determinants in osteosarcoma have included demographics (age, sex), tumour size, site, stage, and the response to chemotherapy. Many of these are determined using varying ...techniques and units of measurement, which can make comparison between studies difficult. The absence of survival difference between limb sparing surgery and amputation has been repeatedly demonstrated in primary disease, and even in the setting of pathological fracture. On the other hand, there is still some controversy over the existence of increased local recurrence for limb-sparing surgery, and the implications of this. Commonly used prognostic determinants such as metastases, and response to chemotherapy enable a high degree of prognostic accuracy but usually at a late stage in the course of disease. Leading on from this, there is a need to uncover molecular pathways with specific influence over osteosarcoma progression to facilitate earlier treatment changes. Some important pathways are already being defined, for example the association of CXCR4 with metastases on presentation, the likelihood of doxorubicin resistance with positive P-glycoprotein, and the reduced survival prediction of over expressed survivin. It is anticipated that the future of osteosarcoma treatment will involve treatment tailored to the molecular profile of tumours at diagnosis, adjuvant therapy directed towards dysfunctional molecular pathways rather than the use of cytotoxics, and a more standardised approach to the measurement of clinical prognostic factors.
Osteosarcoma is the most common primary malignancy of bone. It arises in bone during periods of rapid growth and primarily affects adolescents and young adults. The 5-year survival rate for ...osteosarcoma is 60%–70%, with no significant improvements in prognosis since the advent of multiagent chemotherapy. Diagnosis, staging, and surgical management of osteosarcoma remain focused on our anatomical understanding of the disease. As our knowledge of the molecular pathogenesis of osteosarcoma expands, potential therapeutic targets are being identified. A comprehensive understanding of these mechanisms is essential if we are to improve the prognosis of patients with osteosarcoma through tumour-targeted therapies. This paper will outline the pathogenic mechanisms of osteosarcoma oncogenesis and progression and will discuss some of the more frontline translational studies performed to date in search of novel, safer, and more targeted drugs for disease management.
► Domain knowledge from expert botanical taxonomists can be combined with image processing to aid species identification. ► The shape of leaves, flowers, petals, etc. can be used to identify species ...and define species boundaries. ► Digital images of plants are increasingly available. ► Image processing challenges include object deformations, a large number of classes, and a large numbers of features to be selected.
Plants are of fundamental importance to life on Earth. The shapes of leaves, petals and whole plants are of great significance to plant science, as they can help to distinguish between different species, to measure plant health, and even to model climate change. The growing interest in biodiversity and the increasing availability of digital images combine to make this topic timely. The global shortage of expert taxonomists further increases the demand for software tools that can recognize and characterize plants from images. A robust automated species identification system would allow people with only limited botanical training and expertise to carry out valuable field work.
We review the main computational, morphometric and image processing methods that have been used in recent years to analyze images of plants, introducing readers to relevant botanical concepts along the way. We discuss the measurement of leaf outlines, flower shape, vein structures and leaf textures, and describe a wide range of analytical methods in use. We also discuss a number of systems that apply this research, including prototypes of hand-held digital field guides and various robotic systems used in agriculture. We conclude with a discussion of ongoing work and outstanding problems in the area.
Work scheduling research typically prescribes task sequences implemented by managers. Yet employees often have discretion to deviate from their prescribed sequence. Using data from 2.4 million ...radiological diagnoses, we find that doctors prioritize similar tasks (batching) and those tasks they expect to complete faster (shortest expected processing time). Moreover, they exercise more discretion as they accumulate experience. Exploiting random assignment of tasks to doctors’ queues, instrumental variable models reveal that these deviations erode productivity. This productivity decline lessens as doctors learn from experience. Prioritizing the shortest tasks is particularly detrimental to productivity. Actively grouping similar tasks also reduces productivity, in stark contrast to productivity gains from exogenous grouping, indicating deviation costs outweigh benefits from repetition. By analyzing task completion times, our work highlights the trade-offs between the time required to exercise discretion and the potential gains from doing so, which has implications for how discretion over scheduling should be delegated.
The online appendix is available at
https://doi.org/10.1287/mnsc.2017.2810
.
This paper was accepted by Serguei Netessine, operations management.
Odorants interact with receptors expressed in specialized olfactory neurons, and neurons of the same class send their axons to distinct glomeruli in the brain. The stereotypic spatial glomerular ...activity map generates recognition and the behavioral response for the odorant. The valence of an odorant changes with concentration, typically becoming aversive at higher concentrations. Interestingly, in Drosophila larvae, the odorant (E)-2-hexenal is aversive at low concentrations and attractive at higher concentrations. We investigated the molecular and neural basis of this phenomenon, focusing on how activities of different olfactory neurons conveying opposing effects dictate behaviors. We identified the repellant neuron in the larvae as one expressing the olfactory receptor Or7a, whose activation alone at low concentrations of (E)-2-hexenal elicits an avoidance response in an Or7a-dependent manner. We demonstrate that avoidance can be overcome at higher concentrations by activation of additional neurons that are known to be attractive, most notably odorants that are known activators of Or42a and Or85c. These findings suggest that in the larval stage, the attraction-conveying neurons can overcome the aversion-conveying channels for (E)-2-hexenal.
The quantitative assessment of facial appearance and function is critical in the process of restoring normality and thus minimising morbidity in patients with facial deformities. Three-dimensional ...(3D) scanners have increasingly been applied in clinical settings to circumvent the issues associated with standard approaches, namely, subjectivity. This study aimed to summarise the current literature on the accuracy, reliability, and usability of 3D scanning technologies for soft-tissue facial assessment.
Medline, EMBASE, and Web of Science were searched for studies assessing the accuracy, reliability, and/or clinical usability of 3D scanners in assessing facial morphology. All results were filtered by title, abstract, and finally by full text for relevance.
Eight hundred and thirty-seven results were filtered down to 41 articles that were included in this review. Articles were categorised depending on the 3D visualising principle of the scanner being tested: laser-based scanning, stereophotogrammetry, structured-light scanning, or RGB-D (red, green, blue-depth) sensors.
Of the traditional 3D scanners evaluated in the literature, stereophotogrammetric systems most consistently demonstrate excellent accuracy and reliability in the collection of 3D facial scans. Due to their cost, size, and complexity, these systems are often unsuitable for incorporation into clinical environments with limited availability of resources, space, and time. Recently developed RGB-D sensors can collect accurate static and dynamic 3D facial scans without many of these disadvantages. Still, further improvements in their technical specifications and a greater focus on the development of automated facial assessment software is needed before RGB-D sensors can be universally accepted as a new gold-standard for soft-tissue facial assessment.
PDF
