The Relationship between Color and Human Experience
By Stella Mo, Ascham School
This paper investigates the relationship between colour and human experience. In particular, it explores people’s associations with colour, colour preferences across cultures, and how colours in one’s environment affect emotions and productivity.
Does physical exercise (PE) increase cognitive ability? A person's cognitive ability is linked with their ability to perform mental tasks associated with problem solving and learning (Kiely, 2014). Since cognitive performance is related to memory, many also question if memory related diseases such as Alzheimer's and Dementia could be prevented or treated if there was a way to increase a person’s cognitive ability. Therefore, for decades humans have been trying to figure out ways to improve our cognitive performance and abilities. Humans have been trying to understand how the brain works for centuries, with behavioural neuroscience studies taking form even during the 1700 BC(Columbia University, 2014). Because of my own personal experiences, myself being a professional athlete I have always contemplated if PE would have a positive impact on cognitive ability. Therefore, I set out to perform a meta-analysis on whether or not PE has any impact on a person’s cognitive abilities and what special requirements must be met for PE to influence cognitive ability. For example, if aerobic exercise like running would have the same effect as anaerobic exercise such as stretching. If it is shown that PE improves cognitive abilities, exercise can be a great method of prevention and rehabilitation for people with deteriorating cognitive abilities.
Understanding Automated Emotion Classification for Art with Class Activation Mapping
By Nastasya Anokhina, Crimson Global Academy
Art is one of the most ancient forms of human communication. It is a powerful tool for conveying emotions and information. Over the recent decade, the rapid growth adaptation of AI has resulted in the increased interest in its applications across a number of areas, including visual arts. Likewise, the black-box nature of many AI models has developed the need for their transparency, which generated the field of explainable AI (XAI). Previous research has examined visual arts perception and developed methods to predict emotions evoked by art. To our knowledge, no system exists that utilizes methods to make art emotion classification models interpretable. This work proposes a method to make AI emotion classification systems for art more explainable and compare AI’s perception to that of humans.
Has the COVID-19 Crisis Created an Opportunity for the United Arab Emirates to Lean Towards a Closed Economy
By Muhammad Vakil, Jumeirah College
The COVID-19 pandemic has disrupted the global and domestic supply chains and adversely affected Small and Medium Enterprises (SMEs). It also questioned governments dependence on international trade and finance. This paper adopts a descriptive approach and uses the history of economic crises to evaluate the United Arab Emirates' (UAE) response to the COVID-19. The paper's main policy recommendations are threefold: first, the economic consequences of the COVID-19 have required countries to impose powerful monetary stimuli such as adopting very low-interest-rate policies. However, the results suggest that fiscal policy should be prioritized over monetary policy. Second, the most viable solutions should involve long-term reforms rather than a set of extensive short-term fixes. Moreover, in light of recent anti-globalization trends, some of the closed-economy premises, such as capital control, could make emerging economies more resilient. Traditionally, moving towards a closed economy was considered an obstacle to sustainable economic growth for developing countries.
Heterogeneous Results from Trials of Second-Generation CD19 CAR T-Cells
By Jeremy Nielsen, Haileybury College
CAR T-cell therapies are an increasingly popular treatment modality in relapsed/refractory B-cell malignancies, but responses are highly variable. This review presents results from recent studies investigating CD19-directed CAR T-cell therapies in order to provide an overview of the current therapeutic landscape and identify directions for future research. Relevant studies published between 2015 and 2020 were identified using PubMed and the Cochrane Central Register of Controlled Trials. Database searches identified 2,281 studies, and two additional records were included in the final analysis. Results were too heterogeneous to determine publication bias. This analysis included 21 studies, in which 1245 patients were infused with a CAR T-cell product. The most common diagnosis was non-Hodgkin’s lymphoma (62%, n=762); 576 patients (56%) recorded a complete response. Cytokine release syndrome was the most commonly reported acute toxicity, occurring in 70% (n=853) of patients. Neurological toxicity occurred in 42% (n=477) of patients. Severe CRS and neurological toxicity, defined as grade ≥3, were recorded in 18% (n=217) and 19% (n=204) of patients, respectively. Aggregation of results was complicated by highly heterogeneous outcomes, likely due to inconsistent standards of reporting. To allow for more reliable comparison of CAR T-cell therapies, it is suggested that future investigators implement universal standards regarding dosage, grading of toxicity and assessment of biomarkers.
Convolutional Neural Networks has been around for a few years, and it’s one of the most used neural networks to training data, from simple to complicated one. While Backpropagation (or backward propagation of errors) algorithm has experienced a recent resurgence given the widespread adoption of deep neural networks for image recognition and speech recognition, as it is considered an efficient algorithm, and modern implementations take advantage of specialized GPU to further improve performance. So which one is better to use? In order to find out, we used convolutional neural network and backpropagation (regular) algorithm to train the computer to classify the 70000 images of the Modified National Institute of Standards and Technology (MNIST) dataset. Throughout the test data, we evaluate the efficiency of both neural networks throughout individual variables to find out individual influences of each variables before combining all together.