We consider moment matching techniques for estimation in latent Dirichlet allocation (LDA). By drawing explicit links between LDA and discrete versions of independent component analysis (ICA), we first derive a new set of cumulantbased tensors, with an improved sample complexity. Moreover, we reuse standard ICA techniques such as joint diagonalization of tensors to improve over existing methods...

An optimization based method is proposed to generate moment matching scenarios for numerical integration and its use in stochastic programming. The main advantage of the method is its flexibility: it can generate scenarios matching any prescribed set of moments of the underlying distribution rather than matching all moments up to a certain order, and the distribution can be defined over an arbi...

We investigate a hybrid of two styles of algorithms for deriving bounds for optimization tasks over graphical models: non-iterative message-passing schemes exploiting variable duplication to reduce cluster sizes (e.g. MBE) and iterative methods that re-parameterize the problem’s functions aiming to produce good bounds even if functions are processed independently (e.g. MPLP). In this work we co...

Catastrophic forgetting is a problem of neural networks that loses the information of the first task after training the second task. Here, we propose a method, i.e. incremental moment matching (IMM), to resolve this problem. IMM incrementally matches the moment of the posterior distribution of the neural network which is trained on the first and the second task, respectively. To make the search...

Bayesian learning is often hampered by large computational expense. As a powerful generalization of popular belief propagation, expectation propagation (EP) efficiently approximates the exact Bayesian computation. Nevertheless, EP can be sensitive to outliers and suffer from divergence for difficult cases. To address this issue, we propose a new approximate inference approach, relaxed expectati...

Continuous-time birth-death Markov processes serve as useful models in population biology. When the birth-death rates are nonlinear, the time evolution of the first n order moments of the population is not closed, in the sense that it depends on moments of order higher than n. For analysis purpose, the time evolution of the first n order moments is often made to be closed by approximating these...

Continuous-time birth-death Markov processes serve as useful models in population biology. When the birth-death rates are nonlinear, the time evolution of the first n order moments of the population is not closed, in the sense that it depends on moments of order higher than n. For analysis purposes, the time evolution of the first n order moments is often made to be closed by approximating thes...

A method for fingerprint matching using invariant moment featuresis proposed. The fingerprint image is first preprocessed to enhance the original image by the Short Time Fourier Transform (STFT) analysis. Then, a set of seven invariant moment features is extracted to represent the fingerprint image from a cirque of Interest (COI) based on the reference point of the enhanced fingerprint image. T...

Conditional maximum mean discrepancy (CMMD) can capture the between conditional distributions by drawing support from nonlinear kernel functions; thus, it has been successfully used for pattern classification. However, CMMD does not work well on complex distributions, especially when function fails to correctly characterize difference intraclass similarity and interclass similarity. In this pap...