CVIVApr 3, 2023

Tunable Convolutions with Parametric Multi-Loss Optimization

arXiv:2304.00898v14 citationsh-index: 23
Originality Highly original
AI Analysis

This addresses the need for flexible model control in ill-posed image-to-image translation tasks, offering a drop-in replacement for traditional convolutions with minimal computational overhead.

The paper tackles the problem of tuning neural network behavior at inference time based on external factors like user preferences, by proposing tunable convolutions with parametric multi-loss optimization, which outperforms state-of-the-art control strategies in applications such as image denoising, deblurring, super-resolution, and style transfer.

Behavior of neural networks is irremediably determined by the specific loss and data used during training. However it is often desirable to tune the model at inference time based on external factors such as preferences of the user or dynamic characteristics of the data. This is especially important to balance the perception-distortion trade-off of ill-posed image-to-image translation tasks. In this work, we propose to optimize a parametric tunable convolutional layer, which includes a number of different kernels, using a parametric multi-loss, which includes an equal number of objectives. Our key insight is to use a shared set of parameters to dynamically interpolate both the objectives and the kernels. During training, these parameters are sampled at random to explicitly optimize all possible combinations of objectives and consequently disentangle their effect into the corresponding kernels. During inference, these parameters become interactive inputs of the model hence enabling reliable and consistent control over the model behavior. Extensive experimental results demonstrate that our tunable convolutions effectively work as a drop-in replacement for traditional convolutions in existing neural networks at virtually no extra computational cost, outperforming state-of-the-art control strategies in a wide range of applications; including image denoising, deblurring, super-resolution, and style transfer.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes