Visualizing high-dimensional loss landscapes with Hessian directions

Abstract

Analyzing geometric properties of high-dimensional loss functions, such as local curvature and the existence of other optima around a certain point in loss space, can help provide a better understanding of the interplay between neural network structure, implementation attributes, and learning performance. In this work, we combine concepts from high-dimensional probability and differential geometry to study how curvature properties in lower-dimensional loss representations depend on those in the original loss space. We show that saddle points in the original space are rarely correctly identified as such in expected lower-dimensional representations if random projections are used. The principal curvature in the expected lower-dimensional representation is proportional to the mean curvature in the original loss space. Hence, the mean curvature in the original loss space determines if saddle points appear, on average, as either minima, maxima, or almost flat regions. We use the connection between expected curvature in random projections and mean curvature in the original space (i.e., the normalized Hessian trace) to compute Hutchinson-type trace estimates without calculating Hessian-vector products as in the original Hutchinson method. Because random projections are not suitable to correctly identify saddle information, we propose to study projections along dominant Hessian directions that are associated with the largest and smallest principal curvatures. We connect our findings to the ongoing debate on loss landscape flatness and generalizability. Finally, for different common image classifiers and a function approximator, we show and compare random and Hessian projections of loss landscapes with up to about $7\times 10^6$ parameters.

arXiv comment24 pages, 9 figures

References in corpus (17)

• Visualizing the Loss Landscape of Neural Nets
• The Marginal Value of Adaptive Gradient Methods in Machine Learning
• Train faster, generalize better: Stability of stochastic gradient descent
• New insights and perspectives on the natural gradient method
• Adversarial Weight Perturbation Helps Robust Generalization
• Qualitatively characterizing neural network optimization problems
• On Empirical Comparisons of Optimizers for Deep Learning
• The statistics of critical points of Gaussian fields on large-dimensional spaces
• Implicit energy regularization of neural ordinary-differential-equation control
• Shaping the learning landscape in neural networks around wide flat minima
• On the Efficient Calculation of a Linear Combination of Chi-Square Random Variables with an Application in Counting String Vacua
• Neural Ordinary Differential Equation Control of Dynamics on Graphs
• Entropic gradient descent algorithms and wide flat minima
• Negative eigenvalues of the Hessian in deep neural networks
• Eigenvalue spectra and stability of directed complex networks
• Hessian Eigenspectra of More Realistic Nonlinear Models
• Gradient-free training of neural ODEs for system identification and control using ensemble Kalman inversion

Cited by in corpus (2)

• Unification of Symmetries Inside Neural Networks: Transformer, Feedforward and Neural ODE
• Reconstructing Noisy Gene Regulation Dynamics Using Extrinsic-Noise-Driven Neural Stochastic Differential Equations