Abstract
Traffic flow forecasting is an essential component of an intelligent transportation system to mitigate congestion. Recurrent neural networks, particularly gated recurrent units and long short-term memory, have been the stateof-the-art traffic flow forecasting models for the last few years. However, a more sophisticated and resilient model is necessary to effectively acquire long-range correlations in the time-series data sequence under analysis. The dominant performance of transformers by overcoming the drawbacks of recurrent neural networks in natural language processing might tackle this need and lead to successful time-series forecasting. This article presents a multi-head attention based transformer model for traffic flow forecasting with a comparative analysis between a gated recurrent unit and a long-short term memory-based model on PeMS dataset in this context. The model uses 5 heads with 5 identical layers of encoder and decoder and relies on Square Subsequent Masking techniques. The results demonstrate the promising performance of the transform-based model in predicting long-term traffic flow patterns effectively after feeding it with substantial amount of data. It also demonstrates its worthiness by increasing the mean squared errors and mean absolute percentage errors by (1.25 - 47.8)% and (32.4 - 83.8)%, respectively, concerning the current baselines.

Abstract
Speech recognition aims to convert human speech into text and has applications in security, healthcare, commerce, automobiles, and technology, just to name a few. Inserting residual neural networks before recurrent neural network cells improves accuracy and cuts training time by a good margin. Furthermore, layer normalization instead of batch normalization is more effective in model training and performance enhancement. Also, the size of the datasets presents tremendous influences in achieving the best performance. Leveraging these tricks, this article proposes an automatic speech recognition model with a stacked five layers of customized Residual Convolution Neural Network and seven layers of Bi-Directional Gated Recurrent Units, including a logarithmic so f tmax for the model output. Each of them incorporates a learnable per-element affine parameter-based layer normalization technique. The training and testing of the new model were conducted on the LibriSpeech corpus and LJ Speech dataset. The experimental results demonstrate a character error rate (CER) of 4.7 and 3.61% on the two datasets, respectively, with only 33 million parameters without the requirement of any external language model.

Abstract
Road network studies attracted unprecedented and overwhelming interest in recent years due to the clear relationship between human existence and city evolution. Current studies cover many aspects of a road network, for example, road feature extraction from video/image data, road map generalisation, traffic simulation, optimisation of optimal route finding problems, and traffic state prediction. However, analysing road networks as a complex graph is a field to explore. This study presents comparative studies on the Porto, in Portugal, road network sections, mainly of Matosinhos, Paranhos, and Maia municipalities, regarding degree distributions, clustering coefficients, centrality measures, connected components, k-nearest neighbours, and shortest paths. Further insights into the networks took into account the community structures, page rank, and small-world analysis. The results show that the information exchange efficiency of Matosinhos is 0.8, which is 10 and 12.8% more significant than that of the Maia and Paranhos networks, respectively. Other findings stated are: (1) the studied road networks are very accessible and densely linked; (2) they are small-world in nature, with an average length of the shortest pathways between any two roads of 29.17 units, which as found in the scenario of the Maia road network; and (3) the most critical intersections of the studied network are 'Avenida da Boavista, 4100-119 Porto (latitude: 41.157944, longitude: - 8.629105)', and 'Autoestrada do Norte, Porto (latitude: 41.1687869, longitude: - 8.6400656)', based on the analysis of centrality measures.

Abstract
This paper describes the development of a mechanism to deal with time windows constraints. To the best of our knowledge, the time windows constraints are difficult to be fulfilled even for state-of-the-art methods. Therefore, the main contribution of this paper is to propose a new computational technique to deal with such constraints to ensure the algorithm convergence. We test such technique in two metaheuristics to solve the discrete and dynamic Berth Allocation Problem. A data set generator was created, resulting in a diversity of problems in terms of time windows constraints. A detailed computational analysis was carried out to compare the performance for each metaheuristic.

Abstract
Over the years, methods and algorithms have been extensively studied to solve variations of the rectangular twodimensional strip packing problem (2D-SPP), in which small rectangles must be packed inside a larger object denominated as a strip, while minimizing the space necessary to pack all rectangles. In the rectangular 2D-SPP, constraints are used to restrict the packing process, satisfying physical and real-life practical conditions that can impact the material cutting. The objective of this paper is to present an extensive literature review covering scientific publications about the rectangular 2D-SPP constraints in order to provide a useful foundation to support new research works. A systematic literature review was conducted, and 223 articles were selected and analyzed. Real-life practical constraints concerning the rectangular 2D-SPP were classified into seven different groups. In addition, a bibliometric analysis of the rectangular 2D-SPP academic literature was developed. The most relevant authors, articles, and journals were discussed, and an analysis made concerning the basic constraints (orientation and guillotine cutting) and the main solving methods for the rectangular 2D-SPP. Overall, the present paper indicates opportunities to address real-life practical constraints.

Abstract
A smart grid is the future vision of power systems that will be enabled by artificial intelligence (AI), big data, and the Internet of things (IoT), where digitalization is at the core of the energy sector transformation. However, smart grids require that energy managers become more concerned about the reliability and security of power systems. Therefore, energy planners use various methods and technologies to support the sustainable expansion of power systems, such as electricity demand forecasting models, stochastic optimization, robust optimization, and simulation. Electricity forecasting plays a vital role in supporting the reliable transitioning of power systems. This paper deals with short-term load forecasting (STLF), which has become an active area of research over the last few years, with a handful of studies. STLF deals with predicting demand one hour to 24 h in advance. We extensively experimented with several methodologies from machine learning and a complex case study in Panama. Deep learning is a more advanced learning paradigm in the machine learning field that continues to have significant breakthroughs in domain areas such as electricity forecasting, object detection, speech recognition, etc. We identified that the main predictors of electricity demand in the short term: the previous week's load, the previous day's load, and temperature. We found that the deep learning regression model achieved the best performance, which yielded an R squared (R-2) of 0.93 and a mean absolute percentage error (MAPE) of 2.9%, while the AdaBoost model obtained the worst performance with an R-2 of 0.75 and MAPE of 5.70%.