Filters








6 Hits in 1.1 sec

Detecting Hierarchical Changes in Latent Variable Models [article]

Shintaro Fukushima, Kenji Yamanishi
2020 arXiv   pre-print
The key idea to realize it is to employ the MDL (minimum description length) change statistics for measuring the degree of change, in combination with DNML (decomposed normalized maximum likelihood) code-length  ...  This paper addresses the issue of detecting hierarchical changes in latent variable models (HCDL) from data streams.  ...  DNML Code-Length In order to resolve the problem in the previous section, we employ the decomposed normalized maximum likelihood (DNML) code-length for latent variable models instead of the NML code-length  ... 
arXiv:2011.09465v3 fatcat:c3ztgpckfja4dkg7r45ysb6dbi

Project Dynamics and Emergent Complexity [article]

Christopher M. Schlick, Bruno Demissie
2013 arXiv   pre-print
The formulated VAR and HM models provide the base for the calculation of several closed-form solutions of EMC, solutions that allow an explicit complexity assessment based on the model's independent parameters  ...  According to this principle, EMC is of particular interest for evaluating time-dependent complexity.  ...  Furthermore, Andreas Kräußling, Eric Beutner and Pantaley Dimitrov deserve special acknowledgment for reviewing and correcting the mathematical formulas for the vector autoregression models.  ... 
arXiv:1101.0754v8 fatcat:jp32b2ctabarlhd6k5mh7axxki

Unsupervised Language Acquisition: Theory and Practice [article]

Alexander Clark
2002 arXiv   pre-print
In this thesis I present various algorithms for the unsupervised machine learning of aspects of natural languages using a variety of statistical models.  ...  I carefully examine the interaction between the various components, and show how these algorithms can form the basis for a empiricist model of language acquisition.  ...  Likelihood-based The first class of systems are likelihood-based: they select the maximum likelihood model using a PCFG.  ... 
arXiv:cs/0212024v1 fatcat:lsrnihurufcrrk76ygfmlpk4cu

Methods and Techniques of Complex Systems Science: An Overview [article]

Cosma Rohilla Shalizi (Center for the Study of Complex Systems, University of Michigan)
2006 arXiv   pre-print
These in turn divide, roughly, into tools for analyzing data, tools for constructing and evaluating models, and tools for measuring complexity.  ...  I discuss the principles of statistical learning and model selection; time series analysis; cellular automata; agent-based models; the evaluation of complex-systems models; information theory; and ways  ...  Maximum likelihood estimation minimizes this loss function.  ... 
arXiv:nlin/0307015v4 fatcat:6abwfk4rcjedzfg5my7oghiw74

Cognitive Constructivism and the Epistemic Significance of Sharp Statistical Hypotheses in Natural Sciences [article]

J.M. Stern
2020 arXiv   pre-print
The basic metaphor presented in this text, as a foundation for cognitive constructivism, is that of an eigen-solution, and the verification of its objective epistemic status.  ...  While soundness is the result of "estimation and identification tools", such as ML (maximum likelihood) or MAP (maximum a posteriori) optimization, hypothesis testing and model selection, interpretableness  ...  parameters, latent from manifest variables, etc.  ... 
arXiv:1006.5471v8 fatcat:v375ybolz5bkbe6ordmjnnulhi

Learning inhomogeneous parsimonious Markov models with application to DNA sequence analysis [article]

Ralf Eggeling, Universitäts- Und Landesbibliothek Sachsen-Anhalt, Martin-Luther Universität, Große, Ivo, Prof. Dr., Cerquides Bueno, Jesús, Prof. Dr.
2018
For inferring these models from data, we propose different Bayesian and non-Bayesian learning approaches, both for fully observable data and in the presence of latent variables.  ...  In this work we propose a new class of statistical models that allows modeling complex features in the data while keeping the parameter space small in order to avoid overfitting.  ...  Normalized Maximum Likelihood distribution The two-part code is one approach of MDL model selection, albeit not the only one. Recall that the goal is to select one model ξ from a model class M.  ... 
doi:10.25673/1359 fatcat:btcliz66afcupc3jtcl7pwbxiq