Abstract
To forestall future financial crises, risk governance has been embraced as a line of defense. Therefore, this paper seeks to synthesize the risk governance literature, identifying gaps, and suggesting direction for future research, through a systematic literature review (SLR). Analyzing 151 papers from the Scopus and Web of Science databases, this paper finds a steady increase in academic work on risk governance. Using the theory, context, characteristics, and methodology (TCCM) framework, the study emphasizes the importance of chief risk officers, geographical context coverage, and effectiveness and regulation of risk governance. Methodologically, endogeneity issues are a major concern for researchers, agency theory (AT) being the most popular theory used. Finally, moderating and mediating variables that affect risk governance are identified as important but under-explored. While providing practitioners and policymakers with a framework, empirical testing is encouraged. The study contributes to SDG Goal 8, Target 10 of strengthening financial institutions and promoting a resilient financial system.

Abstract
Multiple Sclerosis is one of the most common diseases of the central nervous system that affects millions of people worldwide. The prediction of this disease is considered a challenge since the symptoms are highly variable as the disease worsens and, as such, it has emerged as a topic that artificial intelligence scientists have tried to challenge. With the goal of providing a brief review that may serve as a starting point for future researchers on such a deep field, this paper puts forward a summary of artificial intelligence applications for Multiple Sclerosis evaluation and diagnosis. It includes a detailed recap of what Multiple Sclerosis is, the connections between artificial intelligence and the human brain, and a description of the main proposals in this field. It also concludes what the most reliable methods are at the present time, discussing approaches that achieve accuracy values up to 98.8%. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Abstract
Machine learning models are widely used in time series forecasting. One way to reduce its computational cost and increase its efficiency is to select only the relevant exogenous features to be fed into the model. With this intention, a study on the feature selection methods: Pearson correlation coefficient, Boruta, Boruta-Shap, IMV-LSTM, and LIME is performed. A new method focused on interpretability, SHAP-LSTM, is proposed, using a deep learning model training process as part of a feature selection algorithm. The methods were compared in 2 different datasets showing comparable results with lesser computational cost when compared with the use of all features. In all datasets, SHAP-LSTM showed competitive results, having comparatively better results on the data with a higher presence of scarce occurring categorical features.

Abstract
Solid research depends on systematic, verifiable and repeatable scientometric analysis. However, scientometric analysis is difficult in the current research landscape characterized by the increasing number of publications per year, intersections between research domains, and the diversity of stakeholders involved in research projects. To address this problem, we propose SciCrowd, a hybrid human-AI mixed-initiative system, which supports the collaboration between Artificial Intelligence services and crowdsourcing services. This work discusses the design and evaluation of SciCrowd. The evaluation is focused on attitudes, concerns and intentions towards use. This study contributes a nuanced understanding of the interplay between algorithmic and human tasks in the process of conducting scientometric analysis.

Abstract
In this paper, we consider a demand response (DR) aggregator responsible for participating in the wholesale electricity market on behalf of the end-users who participated in the DR programs. Thus, the DR aggregator can trade its acquired DR within the short-term electricity markets, i.e., the day-ahead and the balancing (real-time) markets. In the proposed framework, the electricity market prices are considered uncertain, and a robust optimization approach is applied to address the uncertainties to maximize the profit of the DR aggregator. A model for analyzing the impact of the energy storage system (ESS) unit on a DR aggregator's performance is developed to provide more flexibility for the consumers. The direct interactions of a DR aggregator with an ESS are neglected in many models. However, this consideration can lead to improvement in the flexibility of the aggregator and also increase the profit of the entity by trading energy in the short-term markets to charge the ESS during the low-price periods and discharge it to the market while the electricity market prices are high. Hence, it is assumed that the DR aggregator owns an ESS unit and can cover a percentage of its traded power through the ESS. An analysis of the impact of the ESS unit on the DR aggregator's performance is applied to study the most appropriate size of the ESS that can maximize the profit of the aggregator. In addition, renewable energy production is employed for end-users through the installation of rooftop photovoltaic (PV) panels. This demand-side renewable generation can provide more flexibility for the participants in DR programs. Various feasible case studies have been applied to demonstrate the model's effectiveness and usefulness, and conclusions are duly drawn. The numerical results indicate that having an ESS seems necessary when the decision-maker desires to protect its profit from the worst-case scenarios and reduces the negative effect of the uncertain parameter, i.e., the wholesale electricity market prices. Thus, it can be shown that having a greater capacity for the ESS has a significant and direct impact on increasing the profit of the aggregator even in the worst-case scenarios, where the profit rises 20 % when the budget of uncertainty in the robust optimization is equal to 12.

Abstract
https://aimc2023.pubpub.org/pub/9z68g7d2 Music has been commonly recognized as a means of expressing emotions. In this sense, an intense debate emerges from the need to verbalize musical emotions. This concern seems highly relevant today, considering the exponential growth of natural language processing using deep learning models where it is possible to prompt semantic propositions to generate music automatically. This scoping review aims to analyze and discuss the possibilities of music generation conditioned by emotions. To address this topic, we propose a historical perspective that encompasses the different disciplines and methods contributing to this topic. In detail, we review two main paradigms adopted in automatic music generation: rules-based and machine-learning models. Of note are the deep learning architectures that aim to generate high-fidelity music from textual descriptions. These models raise fundamental questions about the expressivity of music, including whether emotions can be represented with words or expressed through them. We conclude that overcoming the limitation and ambiguity of language to express emotions through music, some of the use of deep learning with natural language has the potential to impact the creative industries by providing powerful tools to prompt and generate new musical works.