On the Modularity of Hypernetworks
This work provides theoretical insights into the efficiency of hypernetworks for modular learning tasks, which is incremental but relevant for researchers in neural network design and optimization.
The paper tackles the problem of comparing the modularity of hypernetworks versus embedding-based methods for learning input-dependent functions, showing that under certain conditions, hypernetworks can require orders of magnitude fewer parameters than embedding methods.
In the context of learning to map an input $I$ to a function $h_I:\mathcal{X}\to \mathbb{R}$, two alternative methods are compared: (i) an embedding-based method, which learns a fixed function in which $I$ is encoded as a conditioning signal $e(I)$ and the learned function takes the form $h_I(x) = q(x,e(I))$, and (ii) hypernetworks, in which the weights $θ_I$ of the function $h_I(x) = g(x;θ_I)$ are given by a hypernetwork $f$ as $θ_I=f(I)$. In this paper, we define the property of modularity as the ability to effectively learn a different function for each input instance $I$. For this purpose, we adopt an expressivity perspective of this property and extend the theory of Devore et al. 1996 and provide a lower bound on the complexity (number of trainable parameters) of neural networks as function approximators, by eliminating the requirements for the approximation method to be robust. Our results are then used to compare the complexities of $q$ and $g$, showing that under certain conditions and when letting the functions $e$ and $f$ be as large as we wish, $g$ can be smaller than $q$ by orders of magnitude. This sheds light on the modularity of hypernetworks in comparison with the embedding-based method. Besides, we show that for a structured target function, the overall number of trainable parameters in a hypernetwork is smaller by orders of magnitude than the number of trainable parameters of a standard neural network and an embedding method.