In this paper, we propose an unsupervised neural model for learning a discrete embedding of words. While being discrete, our embedding supports vector arithmetic operations similar to continuous embeddings by interpreting each word as a set of propositional statements describing a rule. The formulation of our vector arithmetic closely reflects the logical structure originating from the symbolic sequential decision making formalism (classical/STRIPS planning). Contrary to the conventional wisdom that discrete representation cannot perform well due to the lack of ability to capture the uncertainty, our representation is competitive against the continuous representations in several downstream tasks. We demonstrate that our embedding is directly compatible with the symbolic, classical planning solvers by performing a "paraphrasing" task. Due to the discrete/logical decision making in classical algorithms with deterministic (non-probabilistic) completeness, and also because it does not require additional training on the paraphrasing dataset, our system can negatively answer a paraphrasing query (inexistence of solutions), and can answer that only some approximate solutions exist -- A feature that is missing in the recent, huge, purely neural language models such as GPT-3.

Designing adequate and precise neural architectures is a challenging task, often done by highly specialized personnel. AutoML is a machine learning field that aims to generate good performing models in an automated way. Spectral data such as those obtained from biological analysis have generally a lot of important information, and these data are specifically well suited to Convolutional Neural Networks (CNN) due to their image-like shape. In this work we present NASirt, an AutoML methodology based on Neural Architecture Search (NAS) that finds high accuracy CNN architectures for spectral datasets. The proposed methodology relies on the Item Response Theory (IRT) for obtaining characteristics from an instance level, such as discrimination and difficulty, and it is able to define a rank of top performing submodels. Several experiments are performed in order to demonstrate the methodology's performance with different spectral datasets. Accuracy results are compared to other benchmarks methods, such as a high performing, manually crafted CNN and the Auto-Keras AutoML tool. The results show that our method performs, in most cases, better than the benchmarks, achieving average accuracy as high as 96.96%.

SVM with an RBF kernel is usually one of the best classification algorithms for most data sets, but it is important to tune the two hyperparameters $C$ and $\gamma$ to the data itself. In general, the selection of the hyperparameters is a non-convex optimization problem and thus many algorithms have been proposed to solve it, among them: grid search, random search, Bayesian optimization, simulated annealing, particle swarm optimization, Nelder Mead, and others. There have also been proposals to decouple the selection of $\gamma$ and $C$. We empirically compare 18 of these proposed search algorithms (with different parameterizations for a total of 47 combinations) on 115 real-life binary data sets. We find (among other things) that trees of Parzen estimators and particle swarm optimization select better hyperparameters with only a slight increase in computation time with respect to a grid search with the same number of evaluations. We also find that spending too much computational effort searching the hyperparameters will not likely result in better performance for future data and that there are no significant differences among the different procedures to select the best set of hyperparameters when more than one is found by the search algorithms.

Explaining recommendations enables users to understand whether recommended items are relevant to their needs and has been shown to increase their trust in the system. More generally, if designing explainable machine learning models is key to check the sanity and robustness of a decision process and improve their efficiency, it however remains a challenge for complex architectures, especially deep neural networks that are often deemed "black-box". In this paper, we propose a novel formulation of interpretable deep neural networks for the attribution task. Differently to popular post-hoc methods, our approach is interpretable by design. Using masked weights, hidden features can be deeply attributed, split into several input-restricted sub-networks and trained as a boosted mixture of experts. Experimental results on synthetic data and real-world recommendation tasks demonstrate that our method enables to build models achieving close predictive performances to their non-interpretable counterparts, while providing informative attribution interpretations.

A burst of technology in the 1960s--the Green Revolution--raised agricultural output significantly across developing economies. Since then, rising incomes have boosted protein consumption worldwide, and elevated new challenges: greenhouse-gas emissions from agriculture are increasing (more than a fifth of all emissions worldwide), while a host of practices, from waste to overfishing, threaten the sustainability of food supplies. The COVID-19 pandemic has brought these concerns to the fore: the disease has disrupted supply chains and demand, perversely increasing the amount of food waste in farms and fields while threatening food security for many. As agriculture gradually regains its footing, participants and stakeholders should be casting an eye ahead, to safeguarding food supplies against the potentially greater and more disruptive effects of climate change. Once again, innovation and advanced technologies could make a powerful contribution to secure and sustainable food production. For example, digital and biotechnologies could improve the health of ruminant livestock, requiring fewer methane-producing animals to meet the world's protein needs. Genetic technologies could play a supporting role by enabling the breeding of animals that produce less methane. Meanwhile, AI and sensors could help food processors sort better and slash waste, and other smart technologies could identify inedible by-products for reprocessing. Data and advanced analytics also could help authorities better monitor and manage the seas to limit overfishing--while enabling boat crews to target and find fish with less effort and waste.

Students' perception of classes measured through their opinions on teaching surveys allows to identify deficiencies and problems, both in the environment and in the learning methodologies. The purpose of this paper is to study, through sentiment analysis using natural language processing (NLP) and machine learning (ML) techniques, those opinions in order to identify topics that are relevant for students, as well as predicting the associated sentiment via polarity analysis. As a result, it is implemented, trained and tested two algorithms to predict the associated sentiment as well as the relevant topics of such opinions. The combination of both approaches then becomes useful to identify specific properties of the students' opinions associated with each sentiment label (positive, negative or neutral opinions) and topic. Furthermore, we explore the possibility that students' perception surveys are carried out without closed questions, relying on the information that students can provide through open questions where they express their opinions about their classes.

This paper presents the D2KLab team's approach to the RecSys Challenge 2019 which focuses on the task of recommending accommodations based on user sessions. What is the feeling of a person who says "Rooms of the hotel are enormous, staff are friendly and efficient"? It is positive. Similarly to the sequence of words in a sentence where one can affirm what the feeling is, analysing a sequence of actions performed by a user in a website can lead to predict what will be the item the user will add to his basket at the end of the shopping session. We propose to use a many-to-one recurrent neural network that learns the probability that a user will click on an accommodation based on the sequence of actions he has performed during his browsing session. More specifically, we combine a rule-based algorithm with a Gated Recurrent Unit RNN in order to sort the list of accommodations that is shown to the user. We optimized the RNN on a validation set, tuning the hyper-parameters such as the learning rate, the batch-size and the accommodation embedding size. This analogy with the sentiment analysis task gives promising results. However, it is computationally demanding in the training phase and it needs to be further tuned.

Benefiting from the rapid development of convolutional neural networks, the performance of car license plate detection and recognition has been largely improved. Nonetheless, challenges still exist especially for real-world applications. In this paper, we present an efficient and accurate framework to solve the license plate detection and recognition tasks simultaneously. It is a lightweight and unified deep neural network, that can be optimized end-to-end and work in real-time. Specifically, for unconstrained scenarios, an anchor-free method is adopted to efficiently detect the bounding box and four corners of a license plate, which are used to extract and rectify the target region features. Then, a novel convolutional neural network branch is designed to further extract features of characters without segmentation. Finally, recognition task is treated as sequence labelling problems, which are solved by Connectionist Temporal Classification (CTC) directly. Several public datasets including images collected from different scenarios under various conditions are chosen for evaluation. A large number of experiments indicate that the proposed method significantly outperforms the previous state-of-the-art methods in both speed and precision.

The importance of algorithmic fairness grows with the increasing impact machine learning has on people's lives. Recent work on fairness metrics shows the need for causal reasoning in fairness constraints. In this work, a practical method named FairTrade is proposed for creating flexible prediction models which integrate fairness constraints on sensitive causal paths. The method uses recent advances in variational inference in order to account for unobserved confounders. Further, a method outline is proposed which uses the causal mechanism estimates to audit black box models. Experiments are conducted on simulated data and on a real dataset in the context of detecting unlawful social welfare. This research aims to contribute to machine learning techniques which honour our ethical and legal boundaries.

'These cool worlds offer the opportunity for new insights into the formation and atmospheres of planets beyond the Solar System,' said paper author and astronomer Aaron Meisner of the National Science Foundation (NSF)'s NOIRLab. 'This collection of cool brown dwarfs also allows us to accurately estimate the number of free-floating worlds roaming interstellar space near the Sun.' Brown dwarves are the'cooling embers' of space -- to small to support the nuclear reactions that power stars, they are faint and challenging to spot, which is why astronomers have been hunting for them close by, in our galactic neighbourhood. Experts believe that brown dwarves cool as they age, starting at near-stellar temperatures but cooling until they are on a par with planets like Earth -- a hypothesis which the recent findings have provided evidence to support. The Backyard Worlds project recruited more than 100,000 citizen scientists to study trillions of pixels of telescope images looking for the subtle signs of planets and brown dwarves moving out in space. According to the astronomers, there is still no substitute for the human eye when it comes to scouring telescope images for subtle evidence of moving objects -- despite recent advances in machine learning and supercomputer hardware. The astronomical data studied was collected by the Nicholas U. Mayall 4-meter Telescope at the Kitt Peak National Observatory in Arizona and the Victor M. Blanco 4-meter Telescope at the Cerro Tololo Inter-American Observatory in Chile. Although the researchers have only published data on the coldest 95 of the finds, the volunteers have identified more than 1,500 brown dwarves in the astronomical data -- a record-breaker for any citizen science program by a factor of 20. 'It's awesome to know that our discoveries are now counted among the Sun's neighbour and will be targets of further research,' said paper co-author and citizen scientist Jim Walla added. The discoveries were part of'Backyard Worlds: Planet 9', a project which recruited more than 100,000 people to scour astronomical data for new'nearby' objects.