A Shooting Formulation of Deep Learning

Abstract

Continuous-depth neural networks can be viewed as deep limits of discrete neural networks whose dynamics resemble a discretization of an ordinary differential equation (ODE). Although important steps have been taken to realize the advantages of such continuous formulations, most current techniques are not truly continuous-depth as they assume identical layers. Indeed, existing works throw into relief the myriad difficulties presented by an infinite-dimensional parameter space in learning a continuous-depth neural ODE. To this end, we introduce a shooting formulation which shifts the perspective from parameterizing a network layer-by-layer to parameterizing over optimal networks described only by a set of initial conditions. For scalability, we propose a novel particle-ensemble parametrization which fully specifies the optimal weight trajectory of the continuous-depth neural network. Our experiments show that our particle-ensemble shooting formulation can achieve competitive performance, especially on long-range forecasting tasks. Finally, though the current work is inspired by continuous-depth neural networks, the particleensemble shooting formulation also applies to discrete-time networks and may lead to a new fertile area of research in deep learning parametrization.

Overcoming Failures of Imagination in AI Infused System Development and Deployment

“…Machine learning for robotics is increasingly growing as a field and has potential of revolutionizing technology in the unprecedented way." (Choromanski et al, 2020) Suggesting the research topic bounds the scope of inquiry (e.g., fairness papers failing to acknowledge limitations or possible unintended negative effects, theory papers suggesting they are exempt from reflections on impact) "[this work] is theoretical and conceptual in nature and so is its likely current broader impact" (Vialard et al, 2020); "[the] study is crucial as it indicates the vulnerability of [DNN] classifiers to adversarial attacks" (Dolatabadi et al, 2020) Emphasizing the net impact of the paper, (e.g. defending, deemphasizing, or 'balancing' harms with unrelated benefits) "the positive impact of foundational research on public datasets, such as is presented in this paper, far outweighs [risks] lying further downstream."…”