We present our work on semi-supervised parsing of natural language sentences, focusing on multi-source crosslingual transfer of delexicalized dependency parsers. We first evaluate the influence of treebank annotation styles on parsing performance, focusing on adposition attachment style. Then, we present KLcpos3 , an empirical language similarity measure, designed and tuned for source parser we...

Generative Adversarial Nets (GANs) have shown promise in image generation and semi-supervised learning (SSL). However, existing GANs in SSL have two problems: (1) the generator and the discriminator (i.e. the classifier) may not be optimal at the same time; and (2) the generator cannot control the semantics of the generated samples. The problems essentially arise from the two-player formulation...

In this paper we evaluate a method for generating synthetic speech at high speaking rates based on the interpolation of hidden semi-Markov models (HSMMs) trained on speech data recorded at normal and fast speaking rates. The subjective evaluation was carried out with both blind listeners, who are used to very fast speaking rates, and sighted listeners. We show that we can achieve a better intel...

This work is devoted to the numerical simulation of the Vlasov equation using a phase space grid. In contrast with Particle In Cell (PIC) methods which are known to be noisy, we propose a semi-Lagrangian type method to discretize the Vlasov equation in the two dimensional phase space. As this kind of method requires a huge computational effort, one has to carry out the simulations on parallel m...

Last lecture concluded with a discussion of semi-random graph models, an interpolation between worst-case analysis and average-case analysis designed to identify robust algorithms in the face of strong impossibility results for worst-case guarantees. This lecture and the next two give three more analysis frameworks that blend aspects of worstand average-case analysis. Today’s model, of self-imp...

We present a semi-Lagrangian method for advection–diffusion and incompressible Navier–Stokes equations. The focus is on constructing stable schemes of secondorder temporal accuracy, as this is a crucial element for the successful application of semi-Lagrangian methods to turbulence simulations. We implement the method in the context of unstructured spectral/hp element discretization, which allo...

In this paper, we propose a novel semi-supervised learning classification framework using box-based smooth ordering and Multiple 1D-embedding-based interpolation method in Ref. 25 for hyperspectral images. Due to the lack of labeled samples, conventional supervised approaches cannot generally perform efficient enough. On the other hand, obtaining labeled samples for hyperspectral image classifi...

It is further believed that any practical attacks against HyRal are not possible with respect to the state-of-the-art. That is, we conclude that HyRal might be resistant to some well-known analysis such as differential attack, higher order differential attack, linear attack(including Truncated Linear Attack), interpolation attacks, algebraic attack (including XL attack and XSL attack), related ...

Motivated by current efforts to construct more realistic spam filtering experimental corpora, we present a newly assembled, publicly available corpus of genuine and unsolicited (spam) email, dubbed GenSpam. We also propose an adaptive model for semi-structured document classification based on language model component interpolation. We compare this with a number of alternative classification mod...

In this article we establish optimal estimates for the first eigenvalue of Schrödinger operators on the d-dimensional unit sphere. These estimates depend on Lp norms of the potential, or of its inverse, and are equivalent to interpolation inequalities on the sphere. We also characterize a semi-classical asymptotic regime and discuss how our estimates on the sphere differ from those on the Eucli...