This paper proposes the definition of core components of the healthcare domain that can be used as basic building blocks for the development of mHealth applications. Using Natural Language Processing techniques, we systematically analyze all health interventions defined in the Nursing Interventions Classification taxonomy to define a comprehensive set of basic components. Our results show that it is feasible to define a finite set of core components from the nursing domain language that can be further composed into health care plans thus establishing a foundation for developing mHealth solutions with reduced technical effort.

Computational learning theory states that many classes of boolean formulas are learnable in polynomial time. This paper addresses the understudied subject of how, in practice, such formulas can be learned by deep neural networks. Specifically, we analyze boolean formulas associated with model-sampling benchmarks, combinatorial optimization problems, and random 3-CNFs with varying degrees of constrainedness. Our experiments indicate that: (i) neural learning generalizes better than pure rule-based systems and pure symbolic approach; (ii) relatively small and shallow neural networks are very good approximators of formulas associated with combinatorial optimization problems; (iii) smaller formulas seem harder to learn, possibly due to the fewer positive (satisfying) examples available; and (iv) interestingly, underconstrained 3-CNF formulas are more challenging to learn than overconstrained ones. Such findings pave the way for a better understanding, construction, and use of neurosymbolic AI methods.

Identifying essential genes and proteins is a critical step towards a better understanding of human biology and pathology. Computational approaches helped to mitigate experimental constraints by exploring machine learning (ML) methods and the correlation of essentiality with biological information, especially protein-protein interaction (PPI) networks, to predict essential genes. Nonetheless, their performance is still limited, as network-based centralities are not exclusive proxies of essentiality, and traditional ML methods are unable to learn from non-euclidean domains such as graphs. Given these limitations, we proposed EPGAT, an approach for Essentiality Prediction based on Graph Attention Networks (GATs), which are attention-based Graph Neural Networks (GNNs), operating on graph-structured data. Our model directly learns gene essentiality patterns from PPI networks, integrating additional evidence from multiomics data encoded as node attributes. We benchmarked EPGAT for four organisms, including humans, accurately predicting gene essentiality with ROC AUC score ranging from 0.78 to 0.97. Our model significantly outperformed network-based and shallow ML-based methods and achieved a very competitive performance against the state-of-the-art node2vec embedding method. Notably, EPGAT was the most robust approach in scenarios with limited and imbalanced training data. Thus, the proposed approach offers a powerful and effective way to identify essential genes and proteins.

In this paper we propose the use of continuous residual modules for graph kernels in Graph Neural Networks. We show how both discrete and continuous residual layers allow for more robust training, being that continuous residual layers are those which are applied by integrating through an Ordinary Differential Equation (ODE) solver to produce their output. We experimentally show that these residuals achieve better results than the ones with non-residual modules when multiple layers are used, mitigating the low-pass filtering effect of GCN-based models. Finally, we apply and analyse the behaviour of these techniques and give pointers to how this technique can be useful in other domains by allowing more predictable behaviour under dynamic times of computation.

Algorithm selection-the mapping of problem instances to algorithms-has been successfully applied to a variety of complex theoretical and practical problems, including computer games. In this paper, we extend the traditional framework, which considers a single decision maker, to adversarial settings, by modeling algorithm selection as a normal-form game. In this “game of algorithm selection,” agents select algorithms to play a computer game on their behalf. The game's payoff matrix stores the relative performance among algorithms. We also consider nonstationary scenarios, where algorithms can learn from previous matches. We apply this approach to real-time strategy game StarCraft, using bots developed for the game as our algorithms. Our experiments suggest that minimax-Q is a suitable method for algorithm selection in both stationary and nonstationary conditions. We proceed by implementing our approach in MegaBot, a fully capable StarCraft bot. MegaBot showed robustness to nonstationarity, competing in the difficult scenario of StarCraft AI tournaments. MegaBot successfully learns how to select algorithms, exhibiting increasing win rates with tournament progress. In 2016, its debut year, it left 60% of opponents behind in two out of three tournaments. In 2017, however, MegaBot faced difficulties as its algorithm portfolio got outdated compared to newer entries.

Large state and action spaces are very challenging to reinforcement learning. However, in many domains there is a set of algorithms available, which estimate the best action given a state. Hence, agents can either directly learn a performance-maximizing mapping from states to actions, or from states to algorithms. We investigate several aspects of this dilemma, showing sufficient conditions for learning over algorithms to outperform over actions for a finite number of training iterations. We present synthetic experiments to further study such systems. Finally, we propose a function approximation approach, demonstrating the effectiveness of learning over algorithms in real-time strategy games.