What occurs in the infinite-width NTK regime?

Question: Under the classical Neural Tangent Kernel infinite-width limit and suitable parameterisation, which description is most accurate? A) The NTK converges to a deterministic limiting kernel and remains approximately constant during gradient-based training B) The network's hidden representations continue to change substantially, while only its final layer behaves like a kernel method C) The network is equivalent to a Gaussian process throughout training because its parameter posterior remains Gaussian D) The NTK becomes the identity matrix, so every training example is learned independently Correct: A Explanation: In the infinite-width NTK regime, the kernel converges to a limiting kernel and changes negligibly during training. The model's function follows kernel gradient dynamics, often described as a lazy-training regime. This is distinct from saying that the Bayesian parameter posterior remains Gaussian. Topic: advanced ML / neural tangent kernel / deep learning theory