Human-sensing applications such as activity recognition, fall detection, and health monitoring have been revolutionized by advancements in artificial intelligence (AI) and machine learning technologies. These applications can significantly impact health management by monitoring human behavior and providing critical data for health assessments. However, due to the variability in individual behaviors, environmental factors, and the physical…
Protein language models (pLMs), trained on protein sequence databases, aim to capture the fitness landscape for property prediction and design tasks. While scaling these models has become common, it assumes that the source databases accurately reflect the fitness landscape, which may not be true. Understanding protein function was historically tied to predicting structure based on…
Multi-agent AI frameworks are essential for addressing the complexities of real-world applications that involve multiple interacting agents. Several challenges include managing and coordinating various AI agents in complex environments, such as ensuring agent autonomy while maintaining a collective goal, facilitating effective communication and coordination among agents, and achieving scalability without compromising performance. Additionally, the framework…
Approximate nearest neighbor search (ANNS) is a critical technology that powers various AI-driven applications such as data mining, search engines, and recommendation systems. The primary objective of ANNS is to identify the closest vectors to a given query in high-dimensional spaces. This process is essential in contexts where finding similar items quickly is crucial, such…
The field of information retrieval has rapidly evolved due to the exponential growth of digital data. With the increasing volume of unstructured data, efficient methods for searching and retrieving relevant information have become more crucial than ever. Traditional keyword-based search techniques often need to capture the nuanced meaning of text, leading to inaccurate or irrelevant…
Self-correction mechanisms have been a significant topic of interest within artificial intelligence, particularly in Large Language Models (LLMs). Self-correction is traditionally seen as a distinctive human trait. Still, researchers have started investigating how it can be applied to LLMs to enhance their capabilities without requiring external inputs. This emerging area explores ways to enable LLMs…
Large Language Models (LLMs) have revolutionized artificial intelligence, impacting various scientific and engineering disciplines. The Transformer architecture, initially designed for machine translation, has become the foundation for GPT models, significantly advancing the field. However, current LLMs face challenges in their training approach, which primarily focuses on predicting the next token based on previous context while…
Large Language Models (LLMs) based on Transformer architectures have revolutionized AI development. However, the complexity of their training process remains poorly understood. A significant challenge in this domain is the inconsistency in optimizer performance. While the Adam optimizer has become the standard for training Transformers, stochastic gradient descent with momentum (SGD), which is highly effective…
Recent research highlights that Transformers, though successful in tasks like arithmetic and algorithms, need help with length generalization, where models handle inputs of unseen lengths. This is crucial for algorithmic tasks such as coding or reasoning, where input length often correlates with problem difficulty. Large language models face this limitation even when scaled due to…
One of the critical challenges in the development and deployment of Large Language Models (LLMs) is ensuring that these models are aligned with human values. As LLMs are applied across diverse fields and tasks, the risk of these models operating in ways that may contradict ethical norms or propagate cultural biases becomes a significant concern.…