Abstract: With the rapid advancement of quantum computing and information technology, cloud-based quantum machine learning has emerged as a promising solution, enabling resource-constrained users to perform quantum machine learning tasks via remote quantum servers while ensuring privacy protection for both data and models. A relatively comprehensive overview of the latest developments in this field is provided, starting from the fundamental theories of quantum inner products and variational quantum algorithms. An analysis is conducted on the implementation details and application examples of various cloud-based quantum machine learning methods based on quantum inner products and cloud-based variational quantum algorithms. Additionally, the challenges faced by current technologies are discussed and insights into future research directions are offered.