Main Problem

Challenges in Domain Knowledge-Integrated AI Research 

• Research combining deep learning and artificial intelligence is referred to by various terms such as 'physics-informed,' 'assisted,' 'hybrid,' 'knowledge-based,' and 'PIML.' However, the scope and effects of each term remain undefined. 

• Particularly, studies focusing on acceleration using artificial intelligence are often included in these categories. However, these studies primarily aim to address the limitations of domain knowledge-based approaches, rather than resolving the shortcomings of artificial intelligence or tackling issues inherent to end-to-end neural networks. 

• Similarly, classification algorithms integrated with machine learning are frequently included in this category. Yet, like other hybrid methods, their purpose does not extend to addressing challenges associated with end-to-end neural networks.