Gaussian processes (GP) are attractive building blocks for many probabilistic models. Their drawbacks, however, are the rapidly increasing inference time and memory requirement alongside increasing data. The problem can be alleviated with compactly supported (CS) covariance functions, which produce sparse covariance matrices that are fast in computations and cheap to store. CS functions have pr...

While automatic speech recognition of either Cantonese or English alone has achieved a great degree of success, recognition of Canton-English code-mixing speech is not as trivial. This paper attempts to analyze the effect of language mixing on recognition performance of code-mixing utterances. By examining the recognition results of Canton-English code-mixing speech, where Canton is the matrix ...

We consider the problem of improving the Gaussian approximate posterior marginals computed by expectation propagation and the Laplace method in latent Gaussian models and propose methods that are similar in spirit to the Laplace approximation of Tierney and Kadane (1986). We show that in the case of sparse Gaussian models, the computational complexity of expectation propagation can be made comp...

Abstract—In this paper, we address the problem of recovering complex-valued signals from a set of complex-valued linear measurements. Approximate message passing (AMP) is one state-ofthe-art algorithm to recover real-valued sparse signals. However, the extension of AMP to complex-valued case is nontrivial and no detailed and rigorous derivation has been explicitly presented. To fill this gap, w...

A novel cyclic-prefix based delay-spread estimation technique for wireless OFDM systems is proposed. In particular, the authors propose a technique for estimating the delays and powers of multipath components when the channel is sparse, i.e. a few strong multipaths distantly spaced in time, and a technique for estimating the RMS delay-spread when the channel has a large number of sample-spaced ...

In supervised learning one wishes to identify a pattern present in a joint distribution P , of instances, label pairs, by providing a function f from instances to labels that has low risk EP `(y, f(x)). To do so, the learner is given access to n iid samples drawn from P . In many real world problems clean samples are not available. Rather, the learner is given access to samples from a corrupted...

Angle-of-arrival estimation is a widely used localisation method. However, multipath propagation leads to decreased performance. In this paper an expression for the CramerRao lower bound for angle-of-arrival estimation with a known transmitted signal in a two path environment is derived. The influence of array structure on the achievable accuracy and the capability of signal separation is inves...

Inspired by the increasing interest in self-organizing social opportunistic networks, we investigate the problem of distributed detection of unknown communities in dynamic random graphs. As a formal framework, we consider the dynamic version of the well-studied Planted Bisection Model dyn-G(n, p, q) where the node set [n] of the network is partitioned into two unknown communities and, at every ...

The central question of this paper is: can we enhance the spectral purity of nonlinear circuits by using polyphase multipath circuits? The basic idea behind polyphase multipath circuits is to split the nonlinear circuits into two or more paths and exploit phase differences between these paths to cancel undesired distortion products. It turns out that it is very well possible to use polyphase mu...

We consider the problem of correcting the posterior marginal approximations computed by expectation propagation and Laplace approximation in latent Gaussian models and propose correction methods that are similar in spirit to the Laplace approximation of Tierney and Kadane (1986). We show that in the case of sparse Gaussian models, the computational complexity of expectation propagation can be m...