TinyML, or Tiny Machine Learning, is the application of Machine Learning in microcontrollers. These microcontrollers have limited resources, including little CPU, little RAM, and low power consumption. The TinyML Foundation is the official website of this field. The main goal is to reduce large ML models so that they can be used with microcontrollers that have limited resources. This is particularly popular among Makers. Harvard offers a series of free courses on TinyML, covering the fundamentals, applications, and deployment. Embedded systems using microcontrollers cannot handle large models, so techniques like algorithm compression are used. Examples of operating systems for microcontrollers include FreeRTOS and Mbed OS. Machine Learning algorithms search for patterns in data, and TinyML uses techniques to compress these algorithms to remain effective. Quantization, pruning, and knowledge distillation are some of the algorithm compression techniques used. Tensor Flow Lite is a tool commonly used in TinyML. TinyML has various applications, ranging from DIY projects to industry and the environment. It can be used in maintenance to predict breakages, monitor the environment in real-time, and assist people with disabilities. However, there are risks involved, including accuracy issues and data privacy concerns. Overall, the focus should be on developing technology that
https://www.ikkaro.net/what-tinyml-is/