.
Intersectionality theory is an important component of critical theory today and is of great value for understanding complex personal identities and the operation of power. Based on the theory of intersectionality, this paper discusses the relationship between identity and power in order to better understand the varying degrees of power influence an individual experiences at the intersection of multiple identities. By examining the relevant literature and summarizing the theory, we first review the development of the intersectionality theory, starting with the research of Kimberly Krenshaw, and analyze his contributions to the analysis of identities in terms of gender, race, class, etc. The study suggests that identity is not a fixed label but varies according to the specific social and cultural context, and there are relationships of interweaving, overlapping and interacting among different identities. Power plays a significant role in this, being the basis for the formation of identity and also through which it is realized and strengthened. By analyzing some examples, it shows the multiple oppressities that marginalized people suffer at various identity intersections and how they are maintained by institutionalized power. It also focuses on the power of people themselves and how they resist or change under the given power. The research in this paper helps to further understand the complexity of identity politics and power relations, and has some reference significance for the formulation of relevant policies and the realization of social equity.
This study was research on the validation accuracy of different convolutional neural network (CNN) architectures in identifying the dog’s emotions based on their facial images. The goal is to recognize which model can best verify an animal's emotion. The four CNN models are: a custom-designed Original CNN (OCNN), MobileNetV2, EfficientNetB3, and EfficientNetB7. The experiment used a labeled dataset from Kaggle, which contains 4,000 dog facial images with four emotional states: happy, angry, sad, and relaxed, to train and preprocess the model. The research is using accuracy, precision, recall, and F1-score to evaluate the performance of the model. It was shown that EfficientNetB3 had the highest F1-score of 0.72, showcasing better model complexity and performance. In contrast, OCNN was underfitting, while EfficientNetB7 showed overfitting. This experiment highlights the importance of model choice in emotion classification tasks and provides some new views for the development of emotional computing systems in animal welfare.
Ever since the COVID-19 pandemic, the popularity of e-learning has skyrocketed. A significant challenge facing teachers online is low student engagement, a problem that has been addressed through action recognition techniques. However, these are lagging indicators that only detect after students have already disengaged, allowing little time for teacher intervention. This research develops a novel model that uses action prediction to proactively detect student disengagement. A Sequence-to-Sequence Bidirectional Long Short-Term Memory (Seq2Seq bi-LSTM) model that predicts the future engagement level was built. When presented with only 70 percent of a video it is able to predict the next engagement level with an accuracy of 85 percent. In addition, this paper introduces the first dataset designed for student engagement prediction models. The results demonstrate that action prediction is a better alternative to action recognition for engagement monitoring and should be further researched.
Urban heat islands (UHIs) are an increasingly significant consequence of urbanization and climate change, contributing to elevated temperatures that affect public health, energy consumption, and urban sustainability. Accurate and interpretable prediction of UHI intensity is essential for supporting effective heat mitigation strategies and informed urban planning. This study presents UHIQ, an interactive urban heat island prediction framework that combines a physics-informed linear regression model with an XGBoost machine learning model to estimate annual daytime UHI intensity using vegetation and elevation-related environmental variables. To improve model transparency, the framework incorporates permutation feature importance and sensitivity analysis, enabling users to examine how changes in environmental conditions influence predicted UHI intensity. An interactive dashboard further supports scenario-based exploration by allowing users to compare predictions from both modeling approaches in real time. The machine learning model achieved higher predictive performance than the physics-informed model, while both models consistently identified vegetation as a key factor influencing urban heat intensity. By integrating predictive modeling, explainability, and interactive visualization within a unified framework, UHIQ provides an accessible decision-support tool that can assist researchers, planners, and policymakers in understanding urban heat dynamics and evaluating strategies for more sustainable and climate-resilient cities.
The integration of artificial intelligence (AI) is transforming medical research and the healthcare system with its role in the management of complex and global medical issues such as cancer. AI can improve operational efficiency to reduce the burden on the medical professional. Integration of AI into the flow from early diagnosis to treatment and remission can be beneficial to efficiently manage various diseases and help improve patient life expectancy. Integration of AI, Machine Learning (ML) and Deep Learning (DL) in healthcare to build accurate, efficient, and effective tools to solve complex healthcare challenges can open the path to making healthcare accessible, affordable, and equitable throughout the world. Gene expressions and regulations are the key to cell development, DNA damage repair, genomic integrity, and responses to drug therapies. The TP53 gene, which is also known as the 'Guardian of Genomes', regulates the production of the p53 protein. Various environmental and bodily stresses can trigger the synthesis of mutated p53, a common finding in many types of cancers. Many studies focus on molecular understanding of mutations, genomic networks, and mutation effects. We propose a machine learning (ML) framework to build generic predictive models for classification, sub-typing, and progression of cancers using mutation-based TP53 data. The primary goal of our study is to showcase the potential of using TP53 based data in building ML tools that can be useful for cancer research and diagnosis. One key challenge to achieve this goal is to find a comprehensive TP53 data set which includes important features to train the models and provide useful results. However, most publicly available datasets are curated from previous publications and lack uniformity or correlation. In this work, we use the curated TP53 mutation data set to build and train ML frameworks and models after applying data processing techniques to infer cancer information. Through data augmentation and feature selection, we achieved $>80 \%$ accuracy in both classification and subtyping. The progression model to predict tumor grade and cancer stage achieves an accuracy of around $> 4 8 \%$. Our study identifies gaps in the available data that can be addressed to improve the model results in the future. The data augmentation can result in data leakage, thereby producing overly accurate results, the limitation that are discussed in the paper. Our approach underscores the potential of integrating multiple types of TP53 genomic data with machine learning models to build useful analytical tools for cancer research.
