Timezone: »

 
Poster
On Monotonic Linear Interpolation of Neural Network Parameters
James Lucas · Juhan Bae · Michael Zhang · Stanislav Fort · Richard Zemel · Roger Grosse

Tue Jul 20 09:00 AM -- 11:00 AM (PDT) @ None #None

Linear interpolation between initial neural network parameters and converged parameters after training with stochastic gradient descent (SGD) typically leads to a monotonic decrease in the training objective. This Monotonic Linear Interpolation (MLI) property, first observed by Goodfellow et al. 2014, persists in spite of the non-convex objectives and highly non-linear training dynamics of neural networks. Extending this work, we evaluate several hypotheses for this property that, to our knowledge, have not yet been explored. Using tools from differential geometry, we draw connections between the interpolated paths in function space and the monotonicity of the network --- providing sufficient conditions for the MLI property under mean squared error. While the MLI property holds under various settings (e.g., network architectures and learning problems), we show in practice that networks violating the MLI property can be produced systematically, by encouraging the weights to move far from initialization. The MLI property raises important questions about the loss landscape geometry of neural networks and highlights the need to further study their global properties.

Author Information

James Lucas (University of Toronto and Vector Institute)
Juhan Bae (University of Toronto and Vector Institute)
Michael Zhang (University of Toronto)
Stanislav Fort (Google AI)
Richard Zemel (Vector Institute)
Roger Grosse (University of Toronto and Vector Institute)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors