Deep Linear Networks Dynamics: Low-Rank Biases Induced by Initialization Scale and L2 Regularization
share
For deep linear networks (DLN), various hyperparameters alter the dynamics of training dramatically. We investigate how the rank of the linear map found by gradient descent is affected by (1) the initialization norm and (2) the addition of L_2 regularization on the parameters. For (1), we study two regimes: (1a) the linear/lazy regime, for large norm initialization; (1b) a saddle-to-saddle regime for small initialization norm. In the (1a) setting, the dynamics of a DLN of any depth is similar to that of a standard linear model, without any low-rank bias. In the (1b) setting, we conjecture that throughout training, gradient descent approaches a sequence of saddles, each corresponding to linear maps of increasing rank, until reaching a minimal rank global minimum. We support this conjecture with a partial proof and some numerical experiments. For (2), we show that adding a L_2 regularization on the parameters corresponds to the addition to the cost of a L_p-Schatten (quasi)norm on the linear map with p=2/L (for a depth-L network), leading to a stronger low-rank bias as L grows. The effect of L_2 regularization on the loss surface depends on the depth: for shallow networks, all critical points are either strict saddles or global minima, whereas for deep networks, some local minima appear. We numerically observe that these local minima can generalize better than global ones in some settings.
READ FULL TEXT