Our work represents another step into the detection and prevention of these ever-more present political manipulation efforts. We, therefore, start by focusing on understanding what the state-of-the-art approaches lack -- since the problem remains, this is a fair assumption. We find concerning issues within the current literature and follow a diverging path. Notably, by placing emphasis on using data features that are less susceptible to malicious manipulation and also on looking for high-level approaches that avoid a granularity level that is biased towards easy-to-spot and low impact cases. We designed and implemented a framework -- Twitter Watch -- that performs structured Twitter data collection, applying it to the Portuguese Twittersphere. We investigate a data snapshot taken on May 2020, with around 5 million accounts and over 120 million tweets (this value has since increased to over 175 million). The analyzed time period stretches from August 2019 to May 2020, with a focus on the Portuguese elections of October 6th, 2019. However, the Covid-19 pandemic showed itself in our data, and we also delve into how it affected typical Twitter behavior. We performed three main approaches: content-oriented, metadata-oriented, and network interaction-oriented. We learn that Twitter's suspension patterns are not adequate to the type of political trolling found in the Portuguese Twittersphere -- identified by this work and by an independent peer - nor to fake news posting accounts. We also surmised that the different types of malicious accounts we independently gathered are very similar both in terms of content and interaction, through two distinct analysis, and are simultaneously very distinct from regular accounts.

We investigate the use of spatial interpolation methods for reconstructing the horizontal near-surface wind field given a sparse set of measurements. In particular, random Fourier features is compared to a set of benchmark methods including Kriging and Inverse distance weighting. Random Fourier features is a linear model $\beta(\pmb x) = \sum_{k=1}^K \beta_k e^{i\omega_k \pmb x}$ approximating the velocity field, with frequencies $\omega_k$ randomly sampled and amplitudes $\beta_k$ trained to minimize a loss function. We include a physically motivated divergence penalty term $|\nabla \cdot \beta(\pmb x)|^2$, as well as a penalty on the Sobolev norm. We derive a bound on the generalization error and derive a sampling density that minimizes the bound. Following (arXiv:2007.10683 [math.NA]), we devise an adaptive Metropolis-Hastings algorithm for sampling the frequencies of the optimal distribution. In our experiments, our random Fourier features model outperforms the benchmark models.

We introduce the task of historical text summarisation, where documents in historical forms of a language are summarised in the corresponding modern language. This is a fundamentally important routine to historians and digital humanities researchers but has never been automated. We compile a high-quality gold-standard text summarisation dataset, which consists of historical German and Chinese news from hundreds of years ago summarised in modern German or Chinese. Based on cross-lingual transfer learning techniques, we propose a summarisation model that can be trained even with no cross-lingual (historical to modern) parallel data, and further benchmark it against state-of-the-art algorithms. We report automatic and human evaluations that distinguish the historic to modern language summarisation task from standard cross-lingual summarisation (i.e., modern to modern language), highlight the distinctness and value of our dataset, and demonstrate that our transfer learning approach outperforms standard cross-lingual benchmarks on this task.

Housebound by a pandemic, humanity slowed its emissions of greenhouse gases in 2020. But Earth paid little heed: Temperatures last year tied the modern record, climate scientists reported last week. Overall, the planet was about 1.25°C warmer than in preindustrial times, a trend that puts climate targets in jeopardy, according to jointly reported assessments from NASA, Berkeley Earth, the U.K. Met Office, and the National Oceanic and Atmospheric Administration. The annual update of global surface temperatures—an average of readings from thousands of weather stations and ocean probes—shows 2020 essentially tied records set in 2016. But the years were nothing alike. Temperatures in 2016 were boosted by a strong El Niño, a weather pattern that warms the globe by blocking the rise of cold deep waters in the eastern Pacific Ocean. Last year, however, the Pacific entered La Niña, which has a cooling effect. That La Niña didn't provide more relief is an unwelcome surprise, says Nerilie Abram, a climate scientist at Australian National University. “It makes me worried about how quickly the global warming trend is growing.” The past 6 years are the six warmest on record, but the warming of the atmosphere is unsteady because of its chaotic nature. The ocean, which absorbs more than 90% of the heat from global warming, displays a steadier trend, and here, too, 2020 was a record year. The upper levels of the ocean contained 20 zettajoules (1021 joules) more heat than in 2019, and the rise was double the typical annual increase, scientists reported last week in Advances in Atmospheric Sciences . The subtropical Atlantic Ocean was particularly hot, fueling a record outbreak of hurricanes, says Lijing Cheng, a climate scientist at the Chinese Academy of Sciences's Institute of Atmospheric Physics who led the work. This heat, monitored down to 2000 meters by a fleet of 4000 robotic probes, is spreading deeper into the ocean while also migrating toward the poles. An extreme heat wave struck the northern Pacific, killing marine life. For the first time, warm Atlantic waters were seen penetrating into the Arctic Ocean, melting sea ice from below and reducing its extent nearly to a record low ( Science , 28 August 2020, p. [1043][1]). The warming ocean and melting ice sheets are raising sea levels by 4.8 millimeters per year, and the rate is accelerating ( Science , 20 November 2020, p. [901][2]). On land, 2020 was even more relentless, with temperatures rising 1.96°C above preindustrial levels, a clear record, Berkeley Earth reported. It was the warmest year ever in Asia and Europe and tied for the warmest in South America. Russia was particularly hot, breaking its previous record by 1.2°C, while swaths of Siberia were 7°C warmer than in preindustrial times, leading to large-scale fires and thawing permafrost that caused buildings to founder and set off oil spills ( Science , 7 August 2020, p. [612][3]). “Siberia was crazy,” says Zeke Hausfather, a climate scientist at the Breakthrough Institute and co-author of the Berkeley Earth analysis. “That heat would effectively be impossible without the warming we've seen.” In Australia, record-setting heat and drought fueled catastrophic bushfires at the start of 2020. Fires torched nearly one-quarter of southeastern Australia's forests and destroyed 3000 homes. Climate change was to blame for the country's “Black Summer,” Abram and co-authors concluded in a study published this month in Communications Earth & Environment . Meanwhile, in the United States, unprecedented heat came to the desert Southwest, which is already warming faster than the rest of the country. Phoenix wilted under its hottest summer ever, averaging 36°C. Arizona's Maricopa county, home to Phoenix, is a leader in addressing heat exposure, yet its heat deaths have hit a new record each year since 2016. In 2020, the number approached 300, a jump of some 50% over the previous year, says David Hondula, a climatologist who studies heat mortality at Arizona State University, Tempe. “It was just off the charts in terms of heat.” ![Figure][4] Turning up the heatCREDITS: (GRAPHIC) N. DESAI/ SCIENCE ; (DATA) MET OFFICE; NASA; BERKELEY EARTH; NOAA Although the global economic slowdown of the COVID-19 pandemic cut carbon dioxide (CO2) emissions by some 7%, atmospheric CO2 is long-lived, and warming from previous emissions is preordained. In any case, the drop in emissions is unlikely to last. Later this year, in May, before photosynthesis in the Northern Hemisphere draws down CO2, the U.K. Met Office predicts that levels of atmospheric CO2 will pass 417 parts per million for several weeks, 50% higher than preindustrial levels. Only dramatic action by the world's countries, far beyond existing efforts, can begin to halt this build up, Cheng says. Should the current rate of warming continue, the world will breach the targets set in the Paris climate agreement—limiting warming to 1.5°C or 2°C—by 2035 and 2065, respectively. But Hausfather says it's quite possible that warming, which has largely held steady for the past few decades at 0.19°C per decade, will actually speed up. The rate of warming over the past 14 years is well above the long-term trend. The debate now, he says, is whether that is an omen of an even darker future. [1]: https://www.sciencemag.org/content/369/6507/1043.full [2]: https://www.sciencemag.org/content/370/6519/901.full [3]: https://www.sciencemag.org/content/369/6504/612.full [4]: pending:yes

Deep Learning (DL) is a two-step classification model that consists feature learning, generating feature representations using unsupervised ways and the supervised learning stage at the last step of model using at least two hidden layers on the proposed structures by fully connected layers depending on of the artificial neural networks. The optimization of the predefined classification parameters for the supervised models eases reaching the global optimality with exact zero training error. The autoencoder (AE) models are the highly generalized ways of the unsupervised stages for the DL to define the output weights of the hidden neurons with various representations. As alternatively to the conventional Extreme Learning Machines (ELM) AE, Hessenberg decomposition-based ELM autoencoder (HessELM-AE) is a novel kernel to generate different presentations of the input data within the intended sizes of the models. The aim of the study is analyzing the performance of the novel Deep AE kernel for clinical availability on electroencephalogram (EEG) with stroke patients. The slow cortical potentials (SCP) training in stroke patients during eight neurofeedback sessions were analyzed using Hilbert-Huang Transform. The statistical features of different frequency modulations were fed into the Deep ELM model for generative AE kernels. The novel Deep ELM-AE kernels have discriminated the brain activity with high classification performances for positivity and negativity tasks in stroke patients.

This thesis is a proof of concept for the potential of Variational Auto-Encoder (VAE) on representation learning of real-world Knowledge Graphs (KG). Inspired by successful approaches to the generation of molecular graphs, we evaluate the capabilities of our model, the Relational Graph Variational Auto-Encoder (RGVAE). The impact of the modular hyperparameter choices, encoding through graph convolutions, graph matching and latent space prior, is compared. The RGVAE is first evaluated on link prediction. The mean reciprocal rank (MRR) scores on the two datasets FB15K-237 and WN18RR are compared to the embedding-based model DistMult. A variational DistMult and a RGVAE without latent space prior constraint are implemented as control models. The results show that between different settings, the RGVAE with relaxed latent space, scores highest on both datasets, yet does not outperform the DistMult. Further, we investigate the latent space in a twofold experiment: first, linear interpolation between the latent representation of two triples, then the exploration of each latent dimension in a $95\%$ confidence interval. Both interpolations show that the RGVAE learns to reconstruct the adjacency matrix but fails to disentangle. For the last experiment we introduce a new validation method for the FB15K-237 data set. The relation type-constrains of generated triples are filtered and matched with entity types. The observed rate of valid generated triples is insignificantly higher than the random threshold. All generated and valid triples are unseen. A comparison between different latent space priors, using the $\delta$-VAE method, reveals a decoder collapse. Finally we analyze the limiting factors of our approach compared to molecule generation and propose solutions for the decoder collapse and successful representation learning of multi-relational KGs.

For some scientific questions, empirical data are essential to develop reliable simulation models. These data usually come from different sources with diverse and heterogeneous formats. The design of complex data-driven models is often shaped by the structure of the data available in research projects. Hence, applying such models to other case studies requires either to get similar data or to transform new data to fit the model inputs. It is the case of agent-based models (ABMs) that use advanced data structures such as Geographic Information Systems data. We faced this problem in the LittoSIM-GEN project when generalizing our participatory flooding model (LittoSIM) to new territories. From this experience, we provide a mapping approach to structure, describe, and automatize the integration of geospatial data into ABMs.

This study develops a neural network-based approach for emulating high-resolution modeled precipitation data with comparable statistical properties but at greatly reduced computational cost. The key idea is to use combination of low- and high- resolution simulations to train a neural network to map from the former to the latter. Specifically, we define two types of CNNs, one that stacks variables directly and one that encodes each variable before stacking, and we train each CNN type both with a conventional loss function, such as mean square error (MSE), and with a conditional generative adversarial network (CGAN), for a total of four CNN variants. We compare the four new CNN-derived high-resolution precipitation results with precipitation generated from original high resolution simulations, a bilinear interpolater and the state-of-the-art CNN-based super-resolution (SR) technique. Results show that the SR technique produces results similar to those of the bilinear interpolator with smoother spatial and temporal distributions and smaller data variabilities and extremes than the original high resolution simulations. While the new CNNs trained by MSE generate better results over some regions than the interpolator and SR technique do, their predictions are still not as close as the original high resolution simulations. The CNNs trained by CGAN generate more realistic and physically reasonable results, better capturing not only data variability in time and space but also extremes such as intense and long-lasting storms. The new proposed CNN-based downscaling approach can downscale precipitation from 50~km to 12~km in 14~min for 30~years once the network is trained (training takes 4~hours using 1~GPU), while the conventional dynamical downscaling would take 1~month using 600 CPU cores to generate simulations at the resolution of 12~km over contiguous United States.

We introduce the "Incremental Implicitly-Refined Classi-fication (IIRC)" setup, an extension to the class incremental learning setup where the incoming batches of classes have two granularity levels. i.e., each sample could have a high-level (coarse) label like "bear" and a low-level (fine) label like "polar bear". Only one label is provided at a time, and the model has to figure out the other label if it has already learnfed it. This setup is more aligned with real-life scenarios, where a learner usually interacts with the same family of entities multiple times, discovers more granularity about them, while still trying not to forget previous knowledge. Moreover, this setup enables evaluating models for some important lifelong learning challenges that cannot be easily addressed under the existing setups. These challenges can be motivated by the example "if a model was trained on the class bear in one task and on polar bear in another task, will it forget the concept of bear, will it rightfully infer that a polar bear is still a bear? and will it wrongfully associate the label of polar bear to other breeds of bear?". We develop a standardized benchmark that enables evaluating models on the IIRC setup. We evaluate several state-of-the-art lifelong learning algorithms and highlight their strengths and limitations. For example, distillation-based methods perform relatively well but are prone to incorrectly predicting too many labels per image. We hope that the proposed setup, along with the benchmark, would provide a meaningful problem setting to the practitioners

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. Attacking enemy cruise missiles, fighter jets, helicopters and longer-range high altitude ballistic missiles all present substantial threats to Navy surface ships, especially when multiple attacks arrive simultaneously. By and large, defending against incoming ballistic missiles and air and cruise missiles requires separate defensive systems … until now. A new family of SPY-6 radar systems is being quickly expanded by the U.S. Navy to incorporate a much wider swath of the fleet.