Changes in climate and environmental conditions significantly impact human behavior. Extreme conditions such as hurricanes, wildfires, snowstorms, and cyclones can increase psychological distress, including elevated levels of stress and anxiety, as well as disruptions to routine psychosocial and daily life activities. These climate-related changes also indirectly affect access to food, secure housing, and overall safety, often resulting in social unrest. The effects are disproportionately felt by low socio-economic groups, who are more vulnerable compared to wealthier populations. Social networks data for keywords related to climate, environmental, mental health were collected for our study and studied the bidirectional relationship between climate data and psychosocial health results, emphasizing the quantification of these effects using structured and validated well-being assessment tools, such as the Happiness Index.
The proliferation of social networks has brought significant changes to many areas of human life. Although Artificial Intelligence (AI) and language models have existed since the 1950s, their growth and widespread use have accelerated since OpenAI began offering accessible platforms in 2018. Social networks have become deeply integrated into daily life, allowing individuals to express their feelings and opinions freely. Environmental conditions can influence human behavior, which may, in turn, be reflected in social media posts. Studies from Large Language Models (LLMs) have become popular for understanding and improving user interactions, little research has explored the relationship between social media content and environmental factors. In this work, we tried to handle this gap by applying LLMs to analyze social media posts and investigate the relationship between environmental conditions and online expressions. Specifically, we considered local temperature and humidity as a variable and found a meaningful correlation between environmental conditions, social media activity, and the outcomes derived from LLM analysis.
In edge AI inference, FPGA Overlay Accelerators offer superior performance-area balance but often lack accurate maximum frequency prediction and efficient Design Space Exploration (DSE) due to complex parameterization. To address these limitations, we propose a Bayesian optimization-based DSE framework using the Tree-Structured Parzen Estimator. We introduce a dependency-aware design space modeling (DAM) to prune the parameter space and a custom regression predictor (CREP) for frequency. To accommodate varying computational constraints, we present two implementations: a lightweight framework DEFA and an advanced framework DEFA-ML. DEFA saves computational cost of CREP and exploration runtime for rapid optimization. Conversely, DEFA-ML leverages a complex prediction model and a Large Language Model assistant for superior optimization quality. We validate the framework on the Altera FPGA AI Suite. For frequency prediction, DEFA provides fast and conservative predictions, while DEFA-ML achieves high linearity with a mean absolute percentage error of 4.21%. In throughput optimization, compared with the FPGA AI Suite baseline optimizer, DEFA improves throughput by 30.16% and DEFA-ML improves 35.12%. In area-throughput trade-off optimization, compared with the baseline, DEFA improves 39.05% and DEFA-ML improves 43.11% for single target model. For multiple target models, DEFA improves 40.76% and DEFA-ML improves 47.66%.
Abstract Achieving comprehensive test coverage is essential for preserving product quality in high-volume manufacturing (HVM). In practice, a persistent gap remains between the defects uncovered at system level and those targeted by structural Automatic Test Pattern Generation (ATPG) during HVM screening. This paper introduces a systematic debug methodology that combines optical and physical failure-analysis techniques to bridge that gap. By applying the method we uncovered fundamental shortcomings in ATPG closure, design-for-test (DFT) structures, and the underlying semiconductor process. The resulting insights enabled upstream design and process refinements that dramatically lowered the risk of quality escapes. In response to the identified coverage holes, new test patterns were created, driving the screening flow toward a defect-parts-per-million (DPPM) rate of < 50. Unlike the conventional reliance on aggregated ATPG stuck-at or transition-delay-fault (TDF) metrics, our results show that system-level failures expose hidden sub-IP and cross-clock-domain gaps even when die and top-level subsystem coverage targets are satisfied. We present a closed-loop debug framework that couples hierarchical ATPG coverage audits with targeted pattern updates and optical/physical failure analysis, providing validated diagnoses, root-cause confirmation, and corrective actions in both test strategy and foundry process control.
