# Training Neural Networks as Learning Data-adaptive Kernels: Provable Representation and Approximation Benefits

arXiv: https://arxiv.org/abs/1901.07114 · HTML: https://mlsanity.com/paper/1901.07114/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 21 Jan 2019.

## Built on by (most cited first, 13 of 13)

- [Learning Operators with Coupled Attention](https://mlsanity.com/paper/2201.01032/) (2022)
- [Quantitative CLTs in Deep Neural Networks](https://mlsanity.com/paper/2307.06092/) (2023)
- [Six Lectures on Linearized Neural Networks](https://mlsanity.com/paper/2308.13431/) (2023)
- [Nonparametric Teaching of Implicit Neural Representations](https://mlsanity.com/paper/2405.10531/) (2024)
- [Implicit Bias of Linear RNNs](https://mlsanity.com/paper/2101.07833/) (2021)
- [A view of mini-batch SGD via generating functions: conditions of convergence, phase transitions, benefit from negative momenta](https://mlsanity.com/paper/2206.11124/) (2022)
- [Tight Convergence Rate Bounds for Optimization Under Power Law Spectral Conditions](https://mlsanity.com/paper/2202.00992/) (2022)
- [A Recipe for Global Convergence Guarantee in Deep Neural Networks](https://mlsanity.com/paper/2104.05785/) (2021)
- [Nonparametric Teaching for Graph Property Learners](https://mlsanity.com/paper/2505.14170/) (2025)
- [Nonasymptotic theory for two-layer neural networks: Beyond the bias-variance trade-off](https://mlsanity.com/paper/2106.04795/) (2021)
- [SGD with memory: fundamental properties and stochastic acceleration](https://mlsanity.com/paper/2410.04228/) (2024)
- [Classification and Adversarial examples in an Overparameterized Linear Model: A Signal Processing Perspective](https://mlsanity.com/paper/2109.13215/) (2021)
- [Nonparametric Teaching of Attention Learners](https://mlsanity.com/paper/2602.20461/) (2026)

## Abstract (arXiv)

Consider the problem: given the data pair (x, y) drawn from a population with f_*(x) = E[y | x = x], specify a neural network model and run gradient flow on the weights over time until reaching any stationarity. How does ft, the function computed by the neural network at time t, relate to f_*, in terms of approximation and representation? What are the provable benefits of the adaptive representation by neural networks compared to the pre-specified fixed basis representation in the classical nonparametric literature? We answer the above questions via a dynamic reproducing kernel Hilbert space (RKHS) approach indexed by the training process of neural networks. Firstly, we show that when reaching any local stationarity, gradient flow learns an adaptive RKHS representation and performs the global least-squares projection onto the adaptive RKHS, simultaneously. Secondly, we prove that as the RKHS is data-adaptive and task-specific, the residual for f_* lies in a subspace that is potentially much smaller than the orthogonal complement of the RKHS. The result formalizes the representation and approximation benefits of neural networks. Lastly, we show that the neural network function computed by gradient flow converges to the kernel ridgeless regression with an adaptive kernel, in the limit of vanishing regularization. The adaptive kernel viewpoint provides new angles of studying the approximation, representation, generalization, and optimization advantages of neural networks.