Running AI models on phones, PCs and embedded devices instead of in the cloud, and how models are made small enough.
55 terms
TinyML is the practice of running machine learning inference on microcontrollers that draw milliwatts and hold models in a few hundred kilobytes of memory.
Unstructured pruning removes individual low-importance weights from a neural network, creating sparse models that can shrink storage for Edge AI deployment.
Visual wake words is a tiny on-device vision task that decides whether a person is present in a camera frame, used to wake larger Edge AI…
Weight clustering compresses a neural network by grouping its weights into a few shared values, shrinking models for Edge AI deployment on small devices.
Weight-only quantization stores model weights in 8, 4 or fewer bits while activations stay in higher precision, shrinking AI models for local devices.