More Like this

1. Cross-layer static resource provisioning for dynamic traffic in flexible grid optical networks

   https://doi.org/10.1364/JOCN.404693  

   Xu, Yuxin; Agrell, Erik; Brandt-Pearce, Maite (January 2021, Journal of Optical Communications and Networking)

   Flexible grid networks need rigorous resource planning to avoid network over-dimensioning and resource over-provisioning. The network must provision the hardware and spectrum resources statically, even for dynamic random bandwidth demands, due to the infrastructure of flexible grid networks, hardware limitations, and reconfiguration speed of the control plane. We propose a flexible online–offline probabilistic (FOOP) algorithm for the static spectrum assignment of random bandwidth demands. The FOOP algorithm considers the probabilistic nature of random bandwidth demands and balances hardware and control plane pressures with spectrum assignment efficiency. The FOOP algorithm uses the probabilistic spectrum Gaussian noise (PSGN) model to estimate the physical-layer impairment (PLI) for random bandwidth traffic. Compared to a benchmark spectrum assignment algorithm and a widely applied PLI estimation model, the proposed FOOP algorithm using the PSGN model saves up to 49% of network resources. 

   more » « less
2. Identification of Linear Latent Variable Model with Arbitrary Distribution

   https://doi.org/10.1609/aaai.v36i6.20585  

   Chen, Zhengming; Xie, Feng; Qiao, Jie; Hao, Zhifeng; Zhang, Kun; Cai, Ruichu (June 2022, Proceedings of the AAAI Conference on Artificial Intelligence)

   An important problem across multiple disciplines is to infer and understand meaningful latent variables. One strategy commonly used is to model the measured variables in terms of the latent variables under suitable assumptions on the connectivity from the latents to the measured (known as measurement model). Furthermore, it might be even more interesting to discover the causal relations among the latent variables (known as structural model). Recently, some methods have been proposed to estimate the structural model by assuming that the noise terms in the measured and latent variables are non-Gaussian. However, they are not suitable when some of the noise terms become Gaussian. To bridge this gap, we investigate the problem of identification of the structural model with arbitrary noise distributions. We provide necessary and sufficient condition under which the structural model is identifiable: it is identifiable iff for each pair of adjacent latent variables Lx, Ly, (1) at least one of Lx and Ly has non-Gaussian noise, or (2) at least one of them has a non-Gaussian ancestor and is not d-separated from the non-Gaussian component of this ancestor by the common causes of Lx and Ly. This identifiability result relaxes the non-Gaussianity requirements to only a (hopefully small) subset of variables, and accordingly elegantly extends the application scope of the structural model. Based on the above identifiability result, we further propose a practical algorithm to learn the structural model. We verify the correctness of the identifiability result and the effectiveness of the proposed method through empirical studies. 

   more » « less
3. Heterogeneous Gaussian Mechanism: Preserving Differential Privacy in Deep Learning with Provable Robustness

   Phan, NhatHai; Vu, Minh; Liu, Yang; Jin, Ruoming; Wu, Xintao; Dou, Dejing; Thai, My T. (October 2019, Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence)

   In this paper, we propose a novel Heterogeneous Gaussian Mechanism (HGM) to preserve differential privacy in deep neural networks, with provable robustness against adversarial examples. We first relax the constraint of the privacy budget in the traditional Gaussian Mechanism from (0, 1] to (0, infty), with a new bound of the noise scale to preserve differential privacy. The noise in our mechanism can be arbitrarily redistributed, offering a distinctive ability to address the trade-off between model utility and privacy loss. To derive provable robustness, our HGM is applied to inject Gaussian noise into the first hidden layer. Then, a tighter robustness bound is proposed. Theoretical analysis and thorough evaluations show that our mechanism notably improves the robustness of differentially private deep neural networks, compared with baseline approaches, under a variety of model attacks. 

   more » « less
4. Heterogeneous Gaussian Mechanism: Preserving Differential Privacy in Deep Learning with Provable Robustness

   https://doi.org/10.24963/ijcai.2019/660  

   Phan, NhatHai; Vu, Minh N.; Liu, Yang; Jin, Ruoming; Dou, Dejing; Wu, Xintao; Thai, My T. (August 2019, Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence)

   In this paper, we propose a novel Heterogeneous Gaussian Mechanism (HGM) to preserve differential privacy in deep neural networks, with provable robustness against adversarial examples. We first relax the constraint of the privacy budget in the traditional Gaussian Mechanism from (0, 1] to (0, infty), with a new bound of the noise scale to preserve differential privacy. The noise in our mechanism can be arbitrarily redistributed, offering a distinctive ability to address the trade-off between model utility and privacy loss. To derive provable robustness, our HGM is applied to inject Gaussian noise into the first hidden layer. Then, a tighter robustness bound is proposed. Theoretical analysis and thorough evaluations show that our mechanism notably improves the robustness of differentially private deep neural networks, compared with baseline approaches, under a variety of model attacks. 

   more » « less
5. Optimized communication strategies with binary coherent states over phase noise channels

   https://doi.org/10.1038/s41534-019-0177-4  

   DiMario, M. T.; Kunz, L.; Banaszek, K.; Becerra, F. E. (July 2019, npj Quantum Information)

   Abstract The achievable rate of information transfer in optical communications is determined by the physical properties of the communication channel, such as the intrinsic channel noise. Bosonic phase noise channels, a class of non-Gaussian channels, have emerged as a relevant noise model in quantum information and optical communication. However, while the fundamental limits for communication over Gaussian channels have been extensively studied, the properties of communication over Bosonic phase noise channels are not well understood. Here we propose and demonstrate experimentally the concept of optimized communication strategies for communication over phase noise channels to enhance information transfer beyond what is possible with conventional methods of modulation and detection. Two key ingredients are generalized constellations of coherent states that interpolate between standard on-off keying and binary phase-shift keying formats, and non-Gaussian measurements based on photon number resolving detection of the coherently displaced signal. For a given power constraint and channel noise strength, these novel strategies rely on joint optimization of the input alphabet and the measurement to provide enhanced communication capability over a non-Gaussian channel characterized in terms of the error rate as well as mutual information. 

   more » « less