Nov 7, 2016 · In this paper, we show that regularizing negatively correlated features is an obstacle for effective decorrelation and present OrthoReg, a novel ...

In this paper, we show that regularizing negatively correlated features is an obstacle for effective decorrelation and present. OrthoReg, a novel regularization ...

Feb 6, 2017 · We show that that models regularized with local feature decorrelation have lower overfitting.

It is shown that regularizing negatively correlated features is an obstacle for effective decorrelation and OrthoReg, a novel regularization technique that ...

TL;DR We propose to locally decorrelate the feature weights of CNNs. When the proposed method, which we call OrthoReg, is used to regularize ...

Nov 7, 2016 · In this paper, we show that regularizing negatively correlated features is an obstacle for effective decorrelation and present OrthoReg, a novel ...

Mar 15, 2017 · In this paper, we show that regularizing negatively correlated features is an obstacle for effective decorrelation and present. OrthoReg, a ...

Apr 24, 2017 · Regularizing CNNs with Locally Constrained Decorrelations. International Conference on Learning Representations (ICLR), (GGS Rating A, Class 2).

Regularizing CNNs with Locally Constrained Decorrelations. P. Rodríguez, J. Gonzàlez, G. Cucurull, J. Gonfaus, and F. Roca. ICLR (Poster), OpenReview.net ...

Explore all code implementations available for Regularizing CNNs with Locally Constrained Decorrelations.