Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

This chapter contains sections titled: Historical Review Supervised Multilayer Networks Unsupervised Neural Networks: Kohonen Network Unsupervised Networks: Adaptive Resonance Theory Network Model Validation Summary References Recommended Exercises

Nonlinear System Identification

Review and big data perspectives on robust data mining approaches for industrial process modeling with outliers and missing data

The analysis of power systems with a significant share of renewable generation using probabilistic tools is essential to appropriately consider the impact that the variability and intermittency of the renewable generation has on the grid. This paper provides a critical assessment and classification of the available probabilistic computational methods that have been applied to power system stability assessment (including small and large disturbance angular stability, voltage and frequency stability). A probabilistic analysis framework with a state-of-the-art review of the existing literature in the area is presented comprising of a review of (i) input variable modelling, (ii) computational methods and (iii) presentation techniques for the output/results. The most widely used probabilistic methods in power system studies are presented with their specific application areas, advantages, and disadvantages. The purpose of this overview, classification, and assessment of the existing methods is to identify the most appropriate probabilistic methods to be used in their current forms, or suitably modified, for different types of stability studies of power systems containing renewable generation.

Existing approaches and trends in uncertainty modelling and probabilistic stability analysis of power systems with renewable generation

Data-driven monitoring of multimode continuous processes: A review

A review of the Expectation Maximization algorithm in data-driven process identification

Approaches to robust process identification: A review and tutorial of probabilistic methods

A new approach for modeling and monitoring of the multivariate processes in presence of faulty and missing observations is introduced. It is assumed that operating modes of the process can transit to each other following a Markov chain model. Transition probabilities of the Markov chain are time varying as a function of the scheduling variable. Therefore, the transition probabilities will be able to vary adaptively according to different operating modes. In order to handle the problem of missing observations and unknown operating regimes, the expectation maximization algorithm is used to estimate the parameters. The proposed method is tested on two simulations and one industrial case studies. The industrial case study is the abnormal operating condition diagnosis in the primary separation vessel of oil-sand processes. In comparison to the conventional methods, the proposed method shows superior performance in detection of different operating conditions of the process. © 2014 American Institute of Chemical Engineers AIChE J, 61: 477–493, 2015

Operating condition diagnosis based on HMM with adaptive transition probabilities in presence of missing observations

A novel approach to process operating mode diagnosis using conditional random fields in the presence of missing data

In this paper, a robust approach to process monitoring and diagnosis based on hidden Markov models with time-varying transition probabilities is proposed. We model observations around different process operating modes by different multivariate Student's $t$ -distributions to describe different likelihoods of outliers. Time-varying transition probabilities assist the model in adapting to new operating conditions. The quality of data in each mode, which is usually affected by the percentage of outliers, is treated by assigning the appropriate degree of freedom for the multivariate $t$ distribution adaptively. The method is compared with other available recent techniques in the literature using simulation and experimental studies and shows a superior performance.

Nima Sammaknejad

Papers

A review of the Expectation Maximization algorithm in data-driven process identification

Approaches to robust process identification: A review and tutorial of probabilistic methods

Operating condition diagnosis based on HMM with adaptive transition probabilities in presence of missing observations

A novel approach to process operating mode diagnosis using conditional random fields in the presence of missing data

Robust Diagnosis of Operating Mode Based on Time-Varying Hidden Markov Models