Dalvan Griebler

Objectives: This protocol aims to explore the relationship between lifestyle and brain health biomarkers, assess the effects of a multifaceted intervention, and identify factors linked to brain resilience in middle-aged adults. Methods: Adults aged 40–59 years from Porto Alegre, RS, Brazil without neurodegenerative or severe psychiatric disorders will be included. The study consists of 3 phases: (1) online questionnaires assessing 7 brain health domains (general health, physical activity, cognition, social life, purpose, sleep, and nutrition); (2) clinical and functional assessments through biomarker collection, using instruments such as a P-WALK 2.0 baropodometric platform, a Cybex isokinetic dynamometer, an InBody 770, an OMRON sphygmomanometer, the Short Physical Performance Battery, and the Timed Up and Go test; (3) a 16-week randomized trial with 4 weekly sessions of strength and aerobic training. Post-intervention assessments will be conducted. Statistical analyses will include t-tests, ANOVA, regressions, and normality tests, using IBM SPSS Statistics, with significance set at p < 0.05. The study has been approved by the Institutional Research Ethics Committee (No. 7.323.434) and was funded by the National Council for Scientific and Technological Development (CNPq). Expected results: Brain health biomarkers should improve, especially in individuals with lower cognitive reserve, and we expect to identify modifiable factors associated with neuropsychological decline. Relevance: Midlife is a critical window for targeting brain health due to its relevance in preventing neurodegenerative diseases. This study seeks to support public health strategies promoting cognitive resilience.

O objetivo desse trabalho é avaliar o impacto do ajuste fino na redução da dimensionalidade do embedding de sentença MiniLM L3, para a tarefa de reconhecimento dimensional de emoções na fala, através de uma abordagem bimodal que combina informações acústicas e textuais. O ajuste fino resultou em um aumento de 3x no Coeficiente de Correlação de Concordância para a dimensão de valência.

As more devices include a display, more probable they may run Doom. Various techniques are used to limit the processing required by the game. A parallel can be drawn with small language models, which are being ported to a wide variety of devices, including legacy ones. While less capable than their large-scale counterparts, these models retain the essence of natural language processing. In this work, we map parallels between Doom and language models, focusing on a TinyLlama model trained on the TinyStories dataset ported to the PlayStation Portable, with hyper-parameter value experimentation.