Neural evaluation metrics derived for numerous speech generation tasks have recently attracted great attention. In this paper, we propose SVSNet, the first end-to-end neural network model to assess speaker voice similarity between converted and natural conversion tasks. Unlike most that use hand-crafted features, SVSNet directly takes raw waveform as input more completely utilize information pr...

In this paper, we propose a generative model, Temporal Generative Adversarial Nets (TGAN), which can learn a semantic representation of unlabeled videos, and is capable of generating videos. Unlike existing Generative Adversarial Nets (GAN)-based methods that generate videos with a single generator consisting of 3D deconvolutional layers, our model exploits two different types of generators: a ...

Fine-grained entity typing aims to assign entity mentions in the free text with types arranged in a hierarchical structure. Traditional distant supervision based methods employ a structured data source as a weak supervision and do not need hand-labeled data, but they neglect the label noise in the automatically labeled training corpus. Although recent studies use many features to prune wrong da...

End-to-end QoS has been seldom studied in its inter-domain aspects, particularly within the scope of the global Internet and from an engineering perspective. This paper is intended to be a contribution in this direction. Starting from a utopian model that achieves any kind of engineering operations without any specific constraint, we deduce the open issues and problems to be solved by a viable ...

One of the aims of the SatNEx (Satellite Network of Excellence, http://www.satnex.de) European project is to provide the scientific community that operates on satellite communications with measurement campaign results and evaluation tools, in order to support research activities that investigate problems in satellite communications at different levels of the OSI stack. In fact, the research act...

Penn Discourse Treebank style discourse parsing is a composite task of detecting explicit and non-explicit discourse relations, their connective and argument spans, and assigning a sense to these relations. Due to the composite nature of the task, the end-to-end performance is greatly affected by the error propagation. This paper describes the end-to-end discourse parser for English submitted t...

We develop a framework for incorporating structured graphical models in the encoders of variational autoencoders (VAEs) that allows us to induce interpretable representations through approximate variational inference. This allows us to both perform reasoning (e.g. classification) under the structural constraints of a given graphical model, and use deep generative models to deal with messy, high...

In a global multi-service and multi-provider market, the Internet Service Providers will increasingly need to differentiate in the service quality they offer and base their operation on new, consumer-centric business models. In this paper, we propose an agent-based framework for the Business-toConsumer (B2C) electronic market, comprising the Consumer Agents, Broker Agents and Content Agents, wh...