<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <html> <head> <link rel="stylesheet" href="style.css" type="text/css"> <meta content="text/html; charset=iso-8859-1" http-equiv="Content-Type"> <link rel="Start" href="index.html"> <link rel="next" href="Gpr_interfaces.Sigs.Deriv.Deriv.Optim.SGD.html"> <link rel="Up" href="Gpr_interfaces.Sigs.Deriv.Deriv.Optim.html"> <link title="Index of types" rel=Appendix href="index_types.html"> <link title="Index of exceptions" rel=Appendix href="index_exceptions.html"> <link title="Index of values" rel=Appendix href="index_values.html"> <link title="Index of modules" rel=Appendix href="index_modules.html"> <link title="Index of module types" rel=Appendix href="index_module_types.html"> <link title="Gpr" rel="Chapter" href="Gpr.html"> <link title="Gpr_interfaces" rel="Chapter" href="Gpr_interfaces.html"> <link title="Gpr_utils" rel="Chapter" href="Gpr_utils.html"> <link title="Gpr_block_diag" rel="Chapter" href="Gpr_block_diag.html"> <link title="Gpr_cov_const" rel="Chapter" href="Gpr_cov_const.html"> <link title="Gpr_cov_lin_ard" rel="Chapter" href="Gpr_cov_lin_ard.html"> <link title="Gpr_cov_lin_one" rel="Chapter" href="Gpr_cov_lin_one.html"> <link title="Gpr_cov_se_iso" rel="Chapter" href="Gpr_cov_se_iso.html"> <link title="Gpr_cov_se_fat" rel="Chapter" href="Gpr_cov_se_fat.html"> <link title="Gpr_fitc_gp" rel="Chapter" href="Gpr_fitc_gp.html"> <link title="Gpr_version" rel="Chapter" href="Gpr_version.html"><title>Gpr_interfaces.Sigs.Deriv.Deriv.Optim.Gsl</title> </head> <body> <div class="navbar"> <a class="up" href="Gpr_interfaces.Sigs.Deriv.Deriv.Optim.html" title="Gpr_interfaces.Sigs.Deriv.Deriv.Optim">Up</a> <a class="post" href="Gpr_interfaces.Sigs.Deriv.Deriv.Optim.SGD.html" title="Gpr_interfaces.Sigs.Deriv.Deriv.Optim.SGD">Next</a> </div> <h1>Module <a href="type_Gpr_interfaces.Sigs.Deriv.Deriv.Optim.Gsl.html">Gpr_interfaces.Sigs.Deriv.Deriv.Optim.Gsl</a></h1> <pre><span class="keyword">module</span> Gsl: <code class="code">sig</code> <a href="Gpr_interfaces.Sigs.Deriv.Deriv.Optim.Gsl.html">..</a> <code class="code">end</code></pre><div class="info module top"> Optimization with the GNU Scientific library (GSL)<br> </div> <hr width="100%"> <pre><span id="EXCEPTIONOptim_exception"><span class="keyword">exception</span> Optim_exception</span> <span class="keyword">of</span> <code class="type">exn</code></pre> <div class="info "> <code class="code">Optim_exception exn</code> is raised when an internal exception occurs, e.g. because GSL fails, or because a callback raised it.<br> </div> <pre><span id="VALtrain"><span class="keyword">val</span> train</span> : <code class="type">?step:float -><br> ?tol:float -><br> ?epsabs:float -><br> ?report_trained_model:(iter:int -> <a href="Gpr_interfaces.Sigs.Eval.Trained.html#TYPEt">Gpr_interfaces.Sigs.Eval.Trained.t</a> -> unit) -><br> ?report_gradient_norm:(iter:int -> float -> unit) -><br> ?kernel:Eval.Spec.Kernel.t -><br> ?sigma2:float -><br> ?inducing:Eval.Spec.Inducing.t -><br> ?n_rand_inducing:int -><br> ?learn_sigma2:bool -><br> ?hypers:Gpr_interfaces.Sigs.Deriv.Deriv.Spec.Hyper.t array -><br> inputs:Eval.Spec.Inputs.t -><br> targets:Lacaml.D.vec -> unit -> <a href="Gpr_interfaces.Sigs.Eval.Trained.html#TYPEt">Gpr_interfaces.Sigs.Eval.Trained.t</a></code></pre><div class="info "> <code class="code">train ?step ?tol ?epsabs ?report_trained_model ?report_gradient_norm ?kernel ?sigma2 ?inducing ?n_rand_inducing ?learn_sigma2 ?hypers ~inputs ~targets ()</code> takes the optional initial optimizer step size <code class="code">step</code>, the optimizer line search tolerance <code class="code">tol</code>, the minimum gradient norm <code class="code">epsabs</code> to achieve by the optimizer, callbacks for reporting intermediate results <code class="code">report_trained_model</code> and <code class="code">report_gradient_norm</code>, an optional <code class="code">kernel</code>, noise level <code class="code">sigma2</code>, inducing inputs <code class="code">inducing</code>, number of randomly chosen inducing inputs <code class="code">n_rand_inducing</code>, a flag for whether the noise level should be learnt <code class="code">learn_sigma2</code>, an array of optional hyper parameters <code class="code">hypers</code> which should be optimized, and the <code class="code">inputs</code> and <code class="code">targets</code>.<br> <b>Returns</b> the trained model obtained by evidence maximization (= type II maximum likelihood).<br> </div> <div class="param_info"><code class="code">step</code> : default = <code class="code">1e-1</code></div> <div class="param_info"><code class="code">tol</code> : default = <code class="code">1e-1</code></div> <div class="param_info"><code class="code">epsabs</code> : default = <code class="code">1e-1</code></div> <div class="param_info"><code class="code">report_trained_model</code> : default = ignored</div> <div class="param_info"><code class="code">report_gradient_norm</code> : default = ignored</div> <div class="param_info"><code class="code">kernel</code> : default = default kernel computed from specification</div> <div class="param_info"><code class="code">sigma2</code> : default = target variance</div> <div class="param_info"><code class="code">inducing</code> : default = randomly selected subset of inputs</div> <div class="param_info"><code class="code">n_rand_inducing</code> : default = 10% of inputs, at most 1000</div> <div class="param_info"><code class="code">learn_sigma2</code> : default = <code class="code">true</code></div> <div class="param_info"><code class="code">hypers</code> : default = all hyper parameters</div> </body></html>