We investigate some aspects of the history of antisemitism in France, one of the cradles of modern antisemitism, using diachronic word embeddings. We constructed a large corpus of French books and periodicals issues that contain a keyword related to Jews and performed a diachronic word embedding over the 1789-1914 period. We studied the changes over time in the semantic spaces of 4 target words and performed embedding projections over 6 streams of antisemitic discourse. This allowed us to track the evolution of antisemitic bias in the religious, economic, socio-politic, racial, ethic and conspiratorial domains. Projections show a trend of growing antisemitism, especially in the years starting in the mid-80s and culminating in the Dreyfus affair. Our analysis also allows us to highlight the peculiar adverse bias towards Judaism in the broader context of other religions.

In this work, we combine two different ranking methods together with several other predictors in a joint random forest approach for the scores of soccer matches. The first ranking method is based on the bookmaker consensus, the second ranking method estimates adequate ability parameters that reflect the current strength of the teams best. The proposed combined approach is then applied to the data from the two previous FIFA Women's World Cups 2011 and 2015. Finally, based on the resulting estimates, the FIFA Women's World Cup 2019 is simulated repeatedly and winning probabilities are obtained for all teams. The model clearly favors the defending champion USA before the host France.

We introduce a lifelong language learning setup where a model needs to learn from a stream of text examples without any dataset identifier. We propose an episodic memory model that performs sparse experience replay and local adaptation to mitigate catastrophic forgetting in this setup. Experiments on text classification and question answering demonstrate the complementary benefits of sparse experience replay and local adaptation to allow the model to continuously learn from new datasets. We also show that the space complexity of the episodic memory module can be reduced significantly ( 50-90%) by randomly choosing which examples to store in memory with a minimal decrease in performance. We consider an episodic memory component as a crucial building block of general linguistic intelligence and see our model as a first step in that direction.

The neural machine translation model has suffered from the lack of large-scale parallel corpora. In contrast, we humans can learn multi-lingual translations even without parallel texts by referring our languages to the external world. To mimic such human learning behavior, we employ images as pivots to enable zero-resource translation learning. However, a picture tells a thousand words, which makes multi-lingual sentences pivoted by the same image noisy as mutual translations and thus hinders the translation model learning. In this work, we propose a progressive learning approach for image-pivoted zero-resource machine translation. Since words are less diverse when grounded in the image, we first learn word-level translation with image pivots, and then progress to learn the sentence-level translation by utilizing the learned word translation to suppress noises in image-pivoted multi-lingual sentences. Experimental results on two widely used image-pivot translation datasets, IAPR-TC12 and Multi30k, show that the proposed approach significantly outperforms other state-of-the-art methods.

Besides the benefits for the community of people with cognitive disabilities, it can also become a useful educational tool in the foreign language classroom, where students with a limited knowledge of a target language can rely on it to adapt complex literary works.

According to its creator, 63 per cent of the population in Sub-Saharan Africa do not have access to global markets because of language barriers. "Over 52 native languages in Africa have undergone language death and have no native speakers," said Emmanuel Gabriel, founder of Germany-based OpenBinacle, the creator of OBTranslate, which was launched this month. "In the next five years, we hope to acquire thousands or millions of users to take up translation tasks on OBTranslate." The innovation resulted from an earlier messaging app that was built in 2017 to allow interaction in real-time translation of 26 African languages, but led to inaccurate outputs, Gabriel admitted. "We were very frustrated about the messaging app, and as a result we didn't want to come into the market with a bad product," added Gabriel.

If you think you've got a bad case of the travel bug, get this: Dr. John Halamka travels 400,000 miles a year. Halamka is chief information officer at Harvard's Beth Israel Deaconess Medical Center, a professor at Harvard Medical School, and a practicing emergency physician. In a talk at Singularity University's Exponential Medicine last week, Halamka shared what he sees as the biggest healthcare problems the world is facing, and the most promising technological solutions from a systems perspective. "In traveling 400,000 miles you get to see lots of different cultures and lots of different people," he said. "And the problems are really the same all over the world. Maybe the cultural context is different or the infrastructure is different, but the problems are very similar."

Timely and accurate agricultural impact assessments for droughts are critical for designing appropriate interventions and policy. These assessments are often ad hoc, late, or spatially imprecise, with reporting at the zonal or regional level. This is problematic as we find substantial variability in losses at the village-level, which is missing when reporting at the zonal level. In this paper, we propose a new data fusion method--combining remotely sensed data with agricultural survey data--that might address these limitations. We apply the method to Ethiopia, which is regularly hit by droughts and is a substantial recipient of ad hoc imported food aid.

Until one and a half years ago, Devidas Lonkar from Chakan town of Pune district had to depend on local fertiliser and pesticide sellers to resolve diseases and fungal issues in his crops. Hailing from an agrarian background, the 26-year-old farmer grows sugarcane, cabbage, cauliflower as well as beetroot and groundnuts across a 7-acre plot. "I would describe the symptoms of fungus or disease to the shopkeeper, to which he would then suggest various pesticides and add-ons. It took me a while before realising that these shopkeepers only suggested chemicals with short-lived efficiency that would inevitably bring farmers back to them within a couple of months," he says. "This app ended up saving me a lot of money as well as time. Sitting at home, I can now diagnose plant diseases and have already saved about Rs 1-1.5 lakh in a year that I would otherwise spend on fertilisers," he mentions.

LiDAR has become a standard sensor for autonomous driving applications as they provide highly precise 3D point clouds. LiDAR is also robust for low-light scenarios at night-time or due to shadows where the performance of cameras is degraded. LiDAR perception is gradually becoming mature for algorithms including object detection and SLAM. However, semantic segmentation algorithm remains to be relatively less explored. Motivated by the fact that semantic segmentation is a mature algorithm on image data, we explore sensor fusion based 3D segmentation. To the best of our knowledge, this is the first attempt at RGB and LiDAR based 3D segmentation for autonomous driving. Our main contribution is to convert the RGB image to a polar-grid mapping representation used for LiDAR and design early and mid-level fusion architectures. Additionally, we design a hybrid fusion architecture that combines both fusion algorithms. We evaluate our algorithm on KITTI dataset which provides segmentation annotation for cars, pedestrians and cyclists. We evaluate two state-of-the-art architectures namely SqueezeSeg and PointSeg and improve the mIoU score by 10 % in both cases relative to the LiDAR only baseline.