Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userApproches quantitatives de lanalyse des pr{e}dictions en traduction automatique neuronale (TAN)
57
0
0.0
(
0
)
Added by
Maria Zimina
Publication date
2020
fields
Informatics Engineering
and research's language is
English
Authors
Maria Zimina-Poirot
Ask ChatGPT about the research
No Arabic abstract
As part of a larger project on optimal learning conditions in neural machine translation, we investigate characteristic training phases of translation engines. All our experiments are carried out using OpenNMT-Py: the pre-processing step is implemented using the Europarl training corpus and the INTERSECT corpus is used for validation. Longitudinal analyses of training phases suggest that the progression of translations is not always linear. Following the results of textometric explorations, we identify the importance of the phenomena related to chronological progression, in order to map different processes at work in neural machine translation (NMT).
rate research
Read More
From 2010, the medical transport has become one of the top ten priorities of the risk management plan in France because of the increase in the cost. For social and medico-social institutions (MSI), this cost represents the second after that of the wages. In this context, the project NOMAd aims an overall improvement of the daily transport of people between their home and their (MSI). To this end, we propose the sharing of transport between several ESMS. This mutualization of transport makes possible to gather and optimize routes in a certain geographical area. The challenge is to improve economic performance while maintaining economic, social and environmental goals. From a scientific point of view, the studied problem is called the Time-Consistent-Dial-a-Ride Problem and aims to find a compromise between the objectives of the cost of transport and the consistency of the service. Given the complexity of the problem, we seek, first of all, to solve the problem for half a day. Then we consider the whole week. To solve these problems, we use the Large Neighborhood Search meta-heuristic and a master problem based on the Set Covering Problem.
The granting process of all credit institutions rejects applicants who seem risky regarding the repayment of their debt. A credit score is calculated and associated with a cut-off value beneath which an applicant is rejected. Developing a new score implies having a learning dataset in which the response variable good/bad borrower is known, so that rejects are de facto excluded from the learning process. We first introduce the context and some useful notations. Then we formalize if this particular sampling has consequences on the scores relevance. Finally, we elaborate on methods that use not-financed clients characteristics and conclude that none of these methods are satisfactory in practice using data from Credit Agricole Consumer Finance. ----- Un syst`eme doctroi de credit peut refuser des demandes de pr^et jugees trop risquees. Au sein de ce syst`eme, le score de credit fournit une valeur mesurant un risque de defaut, valeur qui est comparee `a un seuil dacceptabilite. Ce score est construit exclusivement sur des donnees de clients finances, contenant en particulier linformation `bon ou mauvais payeur, alors quil est par la suite applique `a lensemble des demandes. Un tel score est-il statistiquement pertinent ? Dans cette note, nous precisons et formalisons cette question et etudions leffet de labsence des non-finances sur les scores elabores. Nous presentons ensuite des methodes pour reintegrer les non-finances et concluons sur leur inefficacite en pratique, `a partir de donnees issues de Credit Agricole Consumer Finance.
We summarize here a paper published in 2021 in the DOLAP international workshop DOLAP associated with the EDBT and ICDT conferences. We propose goldMEDAL, a generic metadata model for data lakes based on four concepts and a three-level modeling: conceptual, logical and physical.
The aim of this study is to produce a kinematic analysis of movements of the shoulder complex when hemiparetic patients achieve the grasping of weighted objects. We propose to describe the influence of the weight on kinematic characteristics of the gesture and to establish the relevance of several quantitative indexes concerning the quality of the grasping gesture.
As a result of the deep modifications of the French physics and chemistry curricula in upper secondary school during these last three years, the physics department of the Universite Paris Diderot (France) wished to develop a renovation project concerning its methods of teachings. As science education researchers we took part in this project questioning 104 university academics of this department through 23 exploratory interviews and 81 questionnaires. They have been asked about the possible and desirable changes concerning pedagogical methods and organization (with respect to current difficulties met by the students). This study has enabled us to pinpoint elements of their professional identity as teachers. This identity has been explored according to various directions: rules which govern their profession, qualities and skills for the practice, values of the profession, as well as a didactic dimension, dealing specifically with physics teaching. The analysis of the collected data allows us to identify major trends among university academics conceptions about teaching, such as: a good teacher is above all a good physicist, or, teaching has to favor interactions with students. It also allows us to bring to light tensions between what university academics would like to do in their teaching and what they declare to do in practice or think feasible
suggested questions
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
