Sorry, you need to enable JavaScript to visit this website.

Deep neural networks (DNNs) have recently achieved impressive performances on various applications. However, recent researches show that DNNs are vulnerable to adversarial perturbations injected into input samples. In this paper, we investigate a defense method for face verification: a deep residual generative network (ResGN) is learned to clean adversarial perturbations. We propose a novel training framework composed of ResGN, pre-trained VGG-Face network and FaceNet network.

Usage of passive intelligent surface (PIS) is emerging as a low-cost green alternative to massive antenna systems for realizing high energy beamforming (EB) gains. To maximize its realistic utility, we present a novel channel estimation (CE) protocol for PIS-assisted energy transfer (PET) from a multiantenna power beacon (PB) to a single-antenna energy harvesting (EH) user. Noting the practical limitations of PIS and EH user, all computations are carried out at PB having required active components and radio resources.

Backscatter communication (BSC) is emerging as the core technology for pervasive sustainable internet-of-things applications. However, owing to the resource-limitations of passive tags, this work targets at maximizing the achievable sum-backscattered-throughput by jointly optimizing the transceiver (TRX) design at the full-duplex multiantenna reader and backscattering coefficients (BC) at the single antenna tags.

In this paper, we propose a novel framework for hand localization and pose estimation from a single depth image. For hand localization, unlike most existing methods that using heuristic strategies, e.g. color segmentation, we propose Hierarchical Hand location Networks (HHLN) to estimate the hand location from coarse to fine in depth images, which is robust to the complex environment and efficient. It first applied at a low resolution octree of the whole depth image and produce coarse hand region and then constructs the hand region into a high resolution octree for fine location estimation.

Applications of signal processing and control are classically model-based, involving a two-step procedure for modeling and design: first a model is built from given data, and second, the estimated model is used for filtering, estimation, or control. Both steps typically involve optimization problems, but the combination of both is not necessarily optimal, and the modeling step often ignores the ultimate design objective. Recently, data-driven alternatives are receiving attention, which employ a direct approach combining the modeling and design into a single step.

Speaking style conversion (SSC) is the technology of converting natural speech signals from one style to another. In this study, we propose the use of cycle-consistent adversarial networks (CycleGANs) for converting styles with varying vocal effort, and focus on conversion between normal and Lombard styles as a case study of this problem. We propose a parametric approach that uses the Pulse Model in Log domain (PML) vocoder to extract speech features. These features are mapped using the CycleGAN from utterances in the source style to the corresponding features of target speech.

The ageing population has caused a marked increased in the number of people with cognitive decline linked with dementia. Thus, current diagnostic services are overstretched, and there is an urgent need for automating parts of the assessment process. In previous work, we demonstrated how a stratification tool built around an Intelligent Virtual Agent (IVA) eliciting a conversation by asking memory-probing questions, was able to accurately distinguish between people with a neuro-degenerative disorder (ND) and a functional memory disorder (FMD).

The cloud-based speech recognition/API provides developers or enterprises an easy way to create speech-enabled features in their applications. However, sending audios about personal or company internal information to the cloud, raises concerns about the privacy and security issues. The recognition results generated in cloud may also reveal some sensitive information. This paper proposes a deep polynomial network (DPN) that can be applied to the encrypted speech as an acoustic model. It allows clients to send their data in an encrypted form to the cloud to ensure that their data remains confidential, at mean while the DPN can still make frame-level predictions over the encrypted speech and return them in encrypted form. One good property of the DPN is that it can be trained on unencrypted speech features in the traditional way. To keep the cloud away from the raw audio and recognition results, a cloud-local joint decoding framework is also proposed. We demonstrate the effectiveness of model and framework on the Switchboard and Cortana voice assistant tasks with small performance degradation and latency increased comparing with the traditional cloud-based DNNs.
https://ieeexplore.ieee.org/document/8683721

We investigate the use of entropy-regularized optimal transport (EOT) cost in developing generative models to learn implicit distributions. Two generative models are proposed. One uses EOT cost directly in an one-shot optimization problem and the other uses EOT cost iteratively in an adversarial game. The proposed generative models show improved performance over contemporary models on scores of sample based test.