We present DataGrad, a general back-propagation style training procedure for deep neural architectures that uses regularization of a deep Jacobian-based penalty. It can be viewed as a deep extension of the layerwise contractive auto-encoder penalty. More importantly, it unifies previous proposals for adversarial training of deep neural nets – this list includes directly modifying the gradient, ...

Understanding invariance and discrimination properties of hierarchical models is arguably the key to understanding how and why such models, of which the the mammalian visual system is one instance, can lead to good generalization properties and reduce the sample complexity of a given learning task. In this paper we explore invariance to transformation and the role of layerwise embeddings within...

We propose a connectionist-inspired kernel machine model with three key advantages over traditional kernel machines. First, it is capable of learning distributed and hierarchical representations. Second, its performance is highly robust to the choice of kernel function. Third, the solution space is not limited to the span of images of training data in reproducing kernel Hilbert space (RKHS). To...

Many previous proposals for adversarial training of deep neural nets have included directly modifying the gradient, training on a mix of original and adversarial examples, using contractive penalties, and approximately optimizing constrained adversarial objective functions. In this article, we show that these proposals are actually all instances of optimizing a general, regularized objective we...

Abstract– This article describes an investigation into the free vibration of double-walled carbon nanotubes (DWCNTs) using a nonlocal elastic shell model. Eringen’s nonlocal elasticity is implemented to incorporate the scale effect into the Donnell shell model. Also, the van der Waals interaction between the inner and outer nanotubes is taken into account. A new numerical solution method from i...

A two-phase scheme for accurate predictions of interlaminar stresses in laminated plate and shell structures has been addressed in this study. A modified superconvergent patch recovery (MSPR) technique has been utilized to obtain accurate nodal in-plane stresses which are subsequently used with the thickness integration of the three-dimensional equilibrium equations to evaluate the transverse s...