Toward Less Constrained Macro-Neural Architecture Search.

Networks found with neural architecture search (NAS) achieve the state-of-the-art performance in a variety of tasks, out-performing human-designed networks. However, most NAS methods heavily rely on human-defined assumptions that constrain the search: architecture's outer skeletons, number of layers, parameter heuristics, and search spaces. In addition, common search spaces consist of repeatable modules (cells) instead of fully exploring the architecture's search space by designing entire architectures (macro-search). Imposing such constraints requires deep human expertise and restricts the search to predefined settings. In this article, we propose less constrained macro-neural architecture search (LCMNAS), a method that pushes NAS to less constrained search spaces by performing macro-search without relying on predefined heuristics or bounded search spaces. LCMNAS introduces three components for the NAS pipeline: 1) a method that leverages information about well-known architectures to autonomously generate complex search spaces based on weighted directed graphs (WDGs) with hidden properties; 2) an evolutionary search strategy that generates complete architectures from scratch; and 3) a mixed-performance estimation approach that combines information about architectures at the initialization stage and lower fidelity estimates to infer their trainability and capacity to model complex functions. We present experiments in 14 different datasets showing that LCMNAS is capable of generating both cell and macro-based architectures with minimal GPU computation and state-of-the-art results. Moreover, we conduct extensive studies on the importance of different NAS components in both cell and macro-based settings. The code for reproducibility is publicly available at https://github.com/VascoLopes/LCMNAS.

Mathematical Models to Measure the Variability of Nodes and Networks in Team Sports.

Pattern analysis is a widely researched topic in team sports performance analysis, using information theory as a conceptual framework. Bayesian methods are also used in this research field, but the association between these two is being developed. The aim of this paper is to present new mathematical concepts that are based on information and probability theory and can be applied to network analysis in Team Sports. These results are based on the transition matrices of the Markov chain, associated with the adjacency matrices of a network with n nodes and allowing for a more robust analysis of the variability of interactions in team sports. The proposed models refer to individual and collective rates and indexes of total variability between players and teams as well as the overall passing capacity of a network, all of which are demonstrated in the UEFA 2020/2021 Champions League Final.

Real-time 2D-3D door detection and state classification on a low-power device.

In this paper, we propose three methods for door state classification with the goal to improve robot navigation in indoor spaces. These methods were also developed to be used in other areas and applications since they are not limited to door detection as other related works are. Our methods work offline, in low-powered computers as the Jetson Nano, in real-time with the ability to differentiate between open, closed and semi-open doors. We use the 3D object classification, PointNet, real-time semantic segmentation algorithms such as, FastFCN, FC-HarDNet, SegNet and BiSeNet, the object detection algorithm, DetectNet and 2D object classification networks, AlexNet and GoogleNet. We built a 3D and RGB door dataset with images from several indoor environments using a 3D Realsense camera D435. This dataset is freely available online. All methods are analysed taking into account their accuracy and the speed of the algorithm in a low powered computer. We conclude that it is possible to have a door classification algorithm running in real-time on a low-power device.

Disruptive Mood Dysregulation Disorder: A New Diagnostic Approach to Chronic Irritability in Youth.

(Reprinted from the American Journal of Psychiatry 2014; 171:918-924 with permission from American Psychiatric Association Publishing).

Disruptive mood dysregulation disorder: a new diagnostic approach to chronic irritability in youth.

Disruptive mood dysregulation disorder (DMDD), a newcomer to psychiatric nosology, addresses the need for improved classification and treatment of children exhibiting chronic nonepisodic irritability and severe temper outbursts. In recent years, many of these children have been diagnosed with bipolar disorder, despite the lack of distinct mood episodes. This diagnostic practice has raised concerns, in part because of the escalating prescription of atypical antipsychotics. This article provides an overview of the limited literature on DMDD, including its history and relevant studies of assessment and treatment. A case study is included to illustrate key points, including diagnostic issues that clinicians may encounter when considering a diagnosis of DMDD.

Combination use of atomoxetine hydrochloride and olanzapine in the treatment of attention-deficit/hyperactivity disorder with comorbid disruptive behavior disorder in children and adolescents 10-18 years of age.

OBJECTIVE: The aim of this study was to assess the use of atomoxetine and olanzapine in combination to treat attention-deficit/hyperactivity disorder (ADHD) and comorbid disruptive behaviors in children and adolescents 10-18 years of age. METHODS: Eleven subjects ages 10-18 received open-label atomoxetine and olanzapine for a 10 week treatment period. Patients were assessed at baseline, 2 weeks, 4 weeks, 6 weeks, and 10 weeks (posttreatment). ADHD improvement was measured through the ADHD Rating Scale (ADHD-RS) (Investigator and Parent ratings). Aggression was measured through the Modified Overt Aggression Scale (MOAS). RESULTS: The combined use of atomoxetine and olanzapine resulted in statistically significant improvement in ADHD symptoms and overt aggression from baseline to posttreatment. As evidenced by a 33% reduction in symptoms on the ADHD-RS-I and the MOAS, 73% of patients were considered responders to ADHD treatment, whereas 55% responded to treatment for aggression. Both medications were generally well tolerated. Olanzapine treatment was associated with significant weight gain. Patients gained, on average, 3.9 kg. throughout the treatment period. CONCLUSIONS: These data provide initial evidence that combination use of atomoxetine and olanzapine for the treatment of ADHD and comorbid disruptive behaviors was effective in reducing ADHD symptoms and aggressive behavior in a 10 week treatment period.

Linking selected kinematic, anthropometric and hydrodynamic variables to young swimmer performance.

The aim of this study was to develop a structural equation model (i.e., a confirmatory technique that analyzes relationships among observed variables) for young swimmer performance based on selected kinematic, anthropometric and hydrodynamic variables. A total of 114 subjects (73 boys and 41 girls of mean age of 12.31 ± 1.09 years; 47.91 ± 10.81 kg body mass; 156.57 ± 10.90 cm height and Tanner stages 1-2) were evaluated. The variables assessed were the: (i) 100 [m] freestyle performance; (ii) stroke index; (iii) speed fluctuation; (iv) stroke distance; (v) active drag; (vi) arm span and; (vii) hand surface area. All paths were significant (p < .05). However, in deleting the path between the hand surface area and the stroke index, the model goodness-of-fit significantly improved. Swimming performance in young swimmers appeared to be dependent on swimming efficiency (i.e., stroke index), which is determined by the remaining variables assessed, except for the hand surface area. Therefore, young swimmer coaches and practitioners should design training programs with a focus on technical training enhancement (i.e., improving swimming efficiency).