| <!DOCTYPE html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1.0"><meta name="generator" content="rustdoc"><meta name="description" content="Source of the Rust file `/root/.cargo/git/checkouts/incubator-teaclave-crates-c8106113f74feefc/ede1f68/rusty-machine/src/learning/gmm.rs`."><meta name="keywords" content="rust, rustlang, rust-lang"><title>gmm.rs - source</title><link rel="preload" as="font" type="font/woff2" crossorigin href="../../../SourceSerif4-Regular.ttf.woff2"><link rel="preload" as="font" type="font/woff2" crossorigin href="../../../FiraSans-Regular.woff2"><link rel="preload" as="font" type="font/woff2" crossorigin href="../../../FiraSans-Medium.woff2"><link rel="preload" as="font" type="font/woff2" crossorigin href="../../../SourceCodePro-Regular.ttf.woff2"><link rel="preload" as="font" type="font/woff2" crossorigin href="../../../SourceSerif4-Bold.ttf.woff2"><link rel="preload" as="font" type="font/woff2" crossorigin href="../../../SourceCodePro-Semibold.ttf.woff2"><link rel="stylesheet" href="../../../normalize.css"><link rel="stylesheet" href="../../../rustdoc.css" id="mainThemeStyle"><link rel="stylesheet" href="../../../ayu.css" disabled><link rel="stylesheet" href="../../../dark.css" disabled><link rel="stylesheet" href="../../../light.css" id="themeStyle"><script id="default-settings" ></script><script src="../../../storage.js"></script><script defer src="../../../source-script.js"></script><script defer src="../../../source-files.js"></script><script defer src="../../../main.js"></script><noscript><link rel="stylesheet" href="../../../noscript.css"></noscript><link rel="alternate icon" type="image/png" href="../../../favicon-16x16.png"><link rel="alternate icon" type="image/png" href="../../../favicon-32x32.png"><link rel="icon" type="image/svg+xml" href="../../../favicon.svg"></head><body class="rustdoc source"><!--[if lte IE 11]><div class="warning">This old browser is unsupported and will most likely display funky things.</div><![endif]--><nav class="sidebar"><a class="sidebar-logo" href="../../../rusty_machine/index.html"><div class="logo-container"><img class="rust-logo" src="../../../rust-logo.svg" alt="logo"></div></a></nav><main><div class="width-limiter"><nav class="sub"><a class="sub-logo-container" href="../../../rusty_machine/index.html"><img class="rust-logo" src="../../../rust-logo.svg" alt="logo"></a><form class="search-form"><div class="search-container"><span></span><input class="search-input" name="search" autocomplete="off" spellcheck="false" placeholder="Click or press ‘S’ to search, ‘?’ for more options…" type="search"><div id="help-button" title="help" tabindex="-1"><a href="../../../help.html">?</a></div><div id="settings-menu" tabindex="-1"><a href="../../../settings.html" title="settings"><img width="22" height="22" alt="Change settings" src="../../../wheel.svg"></a></div></div></form></nav><section id="main-content" class="content"><div class="example-wrap"><pre class="src-line-numbers"><span id="1">1</span> |
| <span id="2">2</span> |
| <span id="3">3</span> |
| <span id="4">4</span> |
| <span id="5">5</span> |
| <span id="6">6</span> |
| <span id="7">7</span> |
| <span id="8">8</span> |
| <span id="9">9</span> |
| <span id="10">10</span> |
| <span id="11">11</span> |
| <span id="12">12</span> |
| <span id="13">13</span> |
| <span id="14">14</span> |
| <span id="15">15</span> |
| <span id="16">16</span> |
| <span id="17">17</span> |
| <span id="18">18</span> |
| <span id="19">19</span> |
| <span id="20">20</span> |
| <span id="21">21</span> |
| <span id="22">22</span> |
| <span id="23">23</span> |
| <span id="24">24</span> |
| <span id="25">25</span> |
| <span id="26">26</span> |
| <span id="27">27</span> |
| <span id="28">28</span> |
| <span id="29">29</span> |
| <span id="30">30</span> |
| <span id="31">31</span> |
| <span id="32">32</span> |
| <span id="33">33</span> |
| <span id="34">34</span> |
| <span id="35">35</span> |
| <span id="36">36</span> |
| <span id="37">37</span> |
| <span id="38">38</span> |
| <span id="39">39</span> |
| <span id="40">40</span> |
| <span id="41">41</span> |
| <span id="42">42</span> |
| <span id="43">43</span> |
| <span id="44">44</span> |
| <span id="45">45</span> |
| <span id="46">46</span> |
| <span id="47">47</span> |
| <span id="48">48</span> |
| <span id="49">49</span> |
| <span id="50">50</span> |
| <span id="51">51</span> |
| <span id="52">52</span> |
| <span id="53">53</span> |
| <span id="54">54</span> |
| <span id="55">55</span> |
| <span id="56">56</span> |
| <span id="57">57</span> |
| <span id="58">58</span> |
| <span id="59">59</span> |
| <span id="60">60</span> |
| <span id="61">61</span> |
| <span id="62">62</span> |
| <span id="63">63</span> |
| <span id="64">64</span> |
| <span id="65">65</span> |
| <span id="66">66</span> |
| <span id="67">67</span> |
| <span id="68">68</span> |
| <span id="69">69</span> |
| <span id="70">70</span> |
| <span id="71">71</span> |
| <span id="72">72</span> |
| <span id="73">73</span> |
| <span id="74">74</span> |
| <span id="75">75</span> |
| <span id="76">76</span> |
| <span id="77">77</span> |
| <span id="78">78</span> |
| <span id="79">79</span> |
| <span id="80">80</span> |
| <span id="81">81</span> |
| <span id="82">82</span> |
| <span id="83">83</span> |
| <span id="84">84</span> |
| <span id="85">85</span> |
| <span id="86">86</span> |
| <span id="87">87</span> |
| <span id="88">88</span> |
| <span id="89">89</span> |
| <span id="90">90</span> |
| <span id="91">91</span> |
| <span id="92">92</span> |
| <span id="93">93</span> |
| <span id="94">94</span> |
| <span id="95">95</span> |
| <span id="96">96</span> |
| <span id="97">97</span> |
| <span id="98">98</span> |
| <span id="99">99</span> |
| <span id="100">100</span> |
| <span id="101">101</span> |
| <span id="102">102</span> |
| <span id="103">103</span> |
| <span id="104">104</span> |
| <span id="105">105</span> |
| <span id="106">106</span> |
| <span id="107">107</span> |
| <span id="108">108</span> |
| <span id="109">109</span> |
| <span id="110">110</span> |
| <span id="111">111</span> |
| <span id="112">112</span> |
| <span id="113">113</span> |
| <span id="114">114</span> |
| <span id="115">115</span> |
| <span id="116">116</span> |
| <span id="117">117</span> |
| <span id="118">118</span> |
| <span id="119">119</span> |
| <span id="120">120</span> |
| <span id="121">121</span> |
| <span id="122">122</span> |
| <span id="123">123</span> |
| <span id="124">124</span> |
| <span id="125">125</span> |
| <span id="126">126</span> |
| <span id="127">127</span> |
| <span id="128">128</span> |
| <span id="129">129</span> |
| <span id="130">130</span> |
| <span id="131">131</span> |
| <span id="132">132</span> |
| <span id="133">133</span> |
| <span id="134">134</span> |
| <span id="135">135</span> |
| <span id="136">136</span> |
| <span id="137">137</span> |
| <span id="138">138</span> |
| <span id="139">139</span> |
| <span id="140">140</span> |
| <span id="141">141</span> |
| <span id="142">142</span> |
| <span id="143">143</span> |
| <span id="144">144</span> |
| <span id="145">145</span> |
| <span id="146">146</span> |
| <span id="147">147</span> |
| <span id="148">148</span> |
| <span id="149">149</span> |
| <span id="150">150</span> |
| <span id="151">151</span> |
| <span id="152">152</span> |
| <span id="153">153</span> |
| <span id="154">154</span> |
| <span id="155">155</span> |
| <span id="156">156</span> |
| <span id="157">157</span> |
| <span id="158">158</span> |
| <span id="159">159</span> |
| <span id="160">160</span> |
| <span id="161">161</span> |
| <span id="162">162</span> |
| <span id="163">163</span> |
| <span id="164">164</span> |
| <span id="165">165</span> |
| <span id="166">166</span> |
| <span id="167">167</span> |
| <span id="168">168</span> |
| <span id="169">169</span> |
| <span id="170">170</span> |
| <span id="171">171</span> |
| <span id="172">172</span> |
| <span id="173">173</span> |
| <span id="174">174</span> |
| <span id="175">175</span> |
| <span id="176">176</span> |
| <span id="177">177</span> |
| <span id="178">178</span> |
| <span id="179">179</span> |
| <span id="180">180</span> |
| <span id="181">181</span> |
| <span id="182">182</span> |
| <span id="183">183</span> |
| <span id="184">184</span> |
| <span id="185">185</span> |
| <span id="186">186</span> |
| <span id="187">187</span> |
| <span id="188">188</span> |
| <span id="189">189</span> |
| <span id="190">190</span> |
| <span id="191">191</span> |
| <span id="192">192</span> |
| <span id="193">193</span> |
| <span id="194">194</span> |
| <span id="195">195</span> |
| <span id="196">196</span> |
| <span id="197">197</span> |
| <span id="198">198</span> |
| <span id="199">199</span> |
| <span id="200">200</span> |
| <span id="201">201</span> |
| <span id="202">202</span> |
| <span id="203">203</span> |
| <span id="204">204</span> |
| <span id="205">205</span> |
| <span id="206">206</span> |
| <span id="207">207</span> |
| <span id="208">208</span> |
| <span id="209">209</span> |
| <span id="210">210</span> |
| <span id="211">211</span> |
| <span id="212">212</span> |
| <span id="213">213</span> |
| <span id="214">214</span> |
| <span id="215">215</span> |
| <span id="216">216</span> |
| <span id="217">217</span> |
| <span id="218">218</span> |
| <span id="219">219</span> |
| <span id="220">220</span> |
| <span id="221">221</span> |
| <span id="222">222</span> |
| <span id="223">223</span> |
| <span id="224">224</span> |
| <span id="225">225</span> |
| <span id="226">226</span> |
| <span id="227">227</span> |
| <span id="228">228</span> |
| <span id="229">229</span> |
| <span id="230">230</span> |
| <span id="231">231</span> |
| <span id="232">232</span> |
| <span id="233">233</span> |
| <span id="234">234</span> |
| <span id="235">235</span> |
| <span id="236">236</span> |
| <span id="237">237</span> |
| <span id="238">238</span> |
| <span id="239">239</span> |
| <span id="240">240</span> |
| <span id="241">241</span> |
| <span id="242">242</span> |
| <span id="243">243</span> |
| <span id="244">244</span> |
| <span id="245">245</span> |
| <span id="246">246</span> |
| <span id="247">247</span> |
| <span id="248">248</span> |
| <span id="249">249</span> |
| <span id="250">250</span> |
| <span id="251">251</span> |
| <span id="252">252</span> |
| <span id="253">253</span> |
| <span id="254">254</span> |
| <span id="255">255</span> |
| <span id="256">256</span> |
| <span id="257">257</span> |
| <span id="258">258</span> |
| <span id="259">259</span> |
| <span id="260">260</span> |
| <span id="261">261</span> |
| <span id="262">262</span> |
| <span id="263">263</span> |
| <span id="264">264</span> |
| <span id="265">265</span> |
| <span id="266">266</span> |
| <span id="267">267</span> |
| <span id="268">268</span> |
| <span id="269">269</span> |
| <span id="270">270</span> |
| <span id="271">271</span> |
| <span id="272">272</span> |
| <span id="273">273</span> |
| <span id="274">274</span> |
| <span id="275">275</span> |
| <span id="276">276</span> |
| <span id="277">277</span> |
| <span id="278">278</span> |
| <span id="279">279</span> |
| <span id="280">280</span> |
| <span id="281">281</span> |
| <span id="282">282</span> |
| <span id="283">283</span> |
| <span id="284">284</span> |
| <span id="285">285</span> |
| <span id="286">286</span> |
| <span id="287">287</span> |
| <span id="288">288</span> |
| <span id="289">289</span> |
| <span id="290">290</span> |
| <span id="291">291</span> |
| <span id="292">292</span> |
| <span id="293">293</span> |
| <span id="294">294</span> |
| <span id="295">295</span> |
| <span id="296">296</span> |
| <span id="297">297</span> |
| <span id="298">298</span> |
| <span id="299">299</span> |
| <span id="300">300</span> |
| <span id="301">301</span> |
| <span id="302">302</span> |
| <span id="303">303</span> |
| <span id="304">304</span> |
| <span id="305">305</span> |
| <span id="306">306</span> |
| <span id="307">307</span> |
| <span id="308">308</span> |
| <span id="309">309</span> |
| <span id="310">310</span> |
| <span id="311">311</span> |
| <span id="312">312</span> |
| <span id="313">313</span> |
| <span id="314">314</span> |
| <span id="315">315</span> |
| <span id="316">316</span> |
| <span id="317">317</span> |
| <span id="318">318</span> |
| <span id="319">319</span> |
| <span id="320">320</span> |
| <span id="321">321</span> |
| <span id="322">322</span> |
| <span id="323">323</span> |
| <span id="324">324</span> |
| <span id="325">325</span> |
| <span id="326">326</span> |
| <span id="327">327</span> |
| <span id="328">328</span> |
| <span id="329">329</span> |
| <span id="330">330</span> |
| <span id="331">331</span> |
| <span id="332">332</span> |
| <span id="333">333</span> |
| <span id="334">334</span> |
| <span id="335">335</span> |
| <span id="336">336</span> |
| <span id="337">337</span> |
| <span id="338">338</span> |
| <span id="339">339</span> |
| <span id="340">340</span> |
| <span id="341">341</span> |
| <span id="342">342</span> |
| <span id="343">343</span> |
| <span id="344">344</span> |
| <span id="345">345</span> |
| <span id="346">346</span> |
| <span id="347">347</span> |
| <span id="348">348</span> |
| <span id="349">349</span> |
| <span id="350">350</span> |
| <span id="351">351</span> |
| <span id="352">352</span> |
| <span id="353">353</span> |
| <span id="354">354</span> |
| <span id="355">355</span> |
| <span id="356">356</span> |
| <span id="357">357</span> |
| <span id="358">358</span> |
| <span id="359">359</span> |
| <span id="360">360</span> |
| <span id="361">361</span> |
| <span id="362">362</span> |
| <span id="363">363</span> |
| <span id="364">364</span> |
| <span id="365">365</span> |
| <span id="366">366</span> |
| <span id="367">367</span> |
| <span id="368">368</span> |
| <span id="369">369</span> |
| <span id="370">370</span> |
| <span id="371">371</span> |
| <span id="372">372</span> |
| <span id="373">373</span> |
| <span id="374">374</span> |
| <span id="375">375</span> |
| <span id="376">376</span> |
| <span id="377">377</span> |
| <span id="378">378</span> |
| <span id="379">379</span> |
| <span id="380">380</span> |
| <span id="381">381</span> |
| <span id="382">382</span> |
| </pre><pre class="rust"><code><span class="doccomment">//! Gaussian Mixture Models |
| //! |
| //! Provides implementation of GMMs using the EM algorithm. |
| //! |
| //! # Usage |
| //! |
| //! ``` |
| //! use rusty_machine::linalg::Matrix; |
| //! use rusty_machine::learning::gmm::{CovOption, GaussianMixtureModel}; |
| //! use rusty_machine::learning::UnSupModel; |
| //! |
| //! let inputs = Matrix::new(4, 2, vec![1.0, 2.0, -3.0, -3.0, 0.1, 1.5, -5.0, -2.5]); |
| //! let test_inputs = Matrix::new(3, 2, vec![1.0, 2.0, 3.0, 2.9, -4.4, -2.5]); |
| //! |
| //! // Create gmm with k(=2) classes. |
| //! let mut model = GaussianMixtureModel::new(2); |
| //! model.set_max_iters(10); |
| //! model.cov_option = CovOption::Diagonal; |
| //! |
| //! // Where inputs is a Matrix with features in columns. |
| //! model.train(&inputs).unwrap(); |
| //! |
| //! // Print the means and covariances of the GMM |
| //! println!("{:?}", model.means()); |
| //! println!("{:?}", model.covariances()); |
| //! |
| //! // Where test_inputs is a Matrix with features in columns. |
| //! let post_probs = model.predict(&test_inputs).unwrap(); |
| //! |
| //! // Probabilities that each point comes from each Gaussian. |
| //! println!("{:?}", post_probs.data()); |
| //! ``` |
| </span><span class="kw">use </span>linalg::{Matrix, MatrixSlice, Vector, BaseMatrix, BaseMatrixMut, Axes}; |
| <span class="kw">use </span>rulinalg::utils; |
| <span class="kw">use </span>rulinalg::matrix::decomposition::{PartialPivLu}; |
| |
| <span class="kw">use </span>learning::{LearningResult, UnSupModel}; |
| <span class="kw">use </span>learning::toolkit::rand_utils; |
| <span class="kw">use </span>learning::error::{Error, ErrorKind}; |
| |
| <span class="doccomment">/// Covariance options for GMMs. |
| /// |
| /// - Full : The full covariance structure. |
| /// - Regularized : Adds a regularization constant to the covariance diagonal. |
| /// - Diagonal : Only the diagonal covariance structure. |
| </span><span class="attribute">#[derive(Clone, Copy, Debug)] |
| </span><span class="kw">pub enum </span>CovOption { |
| <span class="doccomment">/// The full covariance structure. |
| </span>Full, |
| <span class="doccomment">/// Adds a regularization constant to the covariance diagonal. |
| </span>Regularized(f64), |
| <span class="doccomment">/// Only the diagonal covariance structure. |
| </span>Diagonal, |
| } |
| |
| |
| <span class="doccomment">/// A Gaussian Mixture Model |
| </span><span class="attribute">#[derive(Debug)] |
| </span><span class="kw">pub struct </span>GaussianMixtureModel { |
| comp_count: usize, |
| mix_weights: Vector<f64>, |
| model_means: <span class="prelude-ty">Option</span><Matrix<f64>>, |
| model_covars: <span class="prelude-ty">Option</span><Vec<Matrix<f64>>>, |
| log_lik: f64, |
| max_iters: usize, |
| <span class="doccomment">/// The covariance options for the GMM. |
| </span><span class="kw">pub </span>cov_option: CovOption, |
| } |
| |
| <span class="kw">impl </span>UnSupModel<Matrix<f64>, Matrix<f64>> <span class="kw">for </span>GaussianMixtureModel { |
| <span class="doccomment">/// Train the model using inputs. |
| </span><span class="kw">fn </span>train(<span class="kw-2">&mut </span><span class="self">self</span>, inputs: <span class="kw-2">&</span>Matrix<f64>) -> LearningResult<()> { |
| <span class="kw">let </span>reg_value = <span class="kw">if </span>inputs.rows() > <span class="number">1 </span>{ |
| <span class="number">1f64 </span>/ (inputs.rows() - <span class="number">1</span>) <span class="kw">as </span>f64 |
| } <span class="kw">else </span>{ |
| <span class="kw">return </span><span class="prelude-val">Err</span>(Error::new(ErrorKind::InvalidData, <span class="string">"Only one row of data provided."</span>)); |
| }; |
| |
| <span class="comment">// Initialization: |
| </span><span class="kw">let </span>k = <span class="self">self</span>.comp_count; |
| |
| <span class="self">self</span>.model_covars = { |
| <span class="kw">let </span>cov_mat = <span class="self">self</span>.initialize_covariances(inputs, reg_value)<span class="question-mark">?</span>; |
| <span class="prelude-val">Some</span>(<span class="macro">vec!</span>[cov_mat; k]) |
| }; |
| |
| <span class="kw">let </span>random_rows: Vec<usize> = |
| rand_utils::reservoir_sample(<span class="kw-2">&</span>(<span class="number">0</span>..inputs.rows()).collect::<Vec<usize>>(), k); |
| <span class="self">self</span>.model_means = <span class="prelude-val">Some</span>(inputs.select_rows(<span class="kw-2">&</span>random_rows)); |
| |
| <span class="kw">for _ in </span><span class="number">0</span>..<span class="self">self</span>.max_iters { |
| <span class="kw">let </span>log_lik_0 = <span class="self">self</span>.log_lik; |
| |
| <span class="kw">let </span>(weights, log_lik_1) = <span class="self">self</span>.membership_weights(inputs)<span class="question-mark">?</span>; |
| |
| <span class="kw">if </span>(log_lik_1 - log_lik_0).abs() < <span class="number">1e-15 </span>{ |
| <span class="kw">break</span>; |
| } |
| |
| <span class="self">self</span>.log_lik = log_lik_1; |
| |
| <span class="self">self</span>.update_params(inputs, weights); |
| } |
| |
| <span class="prelude-val">Ok</span>(()) |
| } |
| |
| <span class="doccomment">/// Predict output from inputs. |
| </span><span class="kw">fn </span>predict(<span class="kw-2">&</span><span class="self">self</span>, inputs: <span class="kw-2">&</span>Matrix<f64>) -> LearningResult<Matrix<f64>> { |
| <span class="kw">if let </span>(<span class="kw-2">&</span><span class="prelude-val">Some</span>(<span class="kw">_</span>), <span class="kw-2">&</span><span class="prelude-val">Some</span>(<span class="kw">_</span>)) = (<span class="kw-2">&</span><span class="self">self</span>.model_means, <span class="kw-2">&</span><span class="self">self</span>.model_covars) { |
| <span class="prelude-val">Ok</span>(<span class="self">self</span>.membership_weights(inputs)<span class="question-mark">?</span>.<span class="number">0</span>) |
| } <span class="kw">else </span>{ |
| <span class="prelude-val">Err</span>(Error::new_untrained()) |
| } |
| |
| } |
| } |
| |
| <span class="kw">impl </span>GaussianMixtureModel { |
| <span class="doccomment">/// Constructs a new Gaussian Mixture Model |
| /// |
| /// Defaults to 100 maximum iterations and |
| /// full covariance structure. |
| /// |
| /// # Examples |
| /// ``` |
| /// use rusty_machine::learning::gmm::GaussianMixtureModel; |
| /// |
| /// let gmm = GaussianMixtureModel::new(3); |
| /// ``` |
| </span><span class="kw">pub fn </span>new(k: usize) -> GaussianMixtureModel { |
| GaussianMixtureModel { |
| comp_count: k, |
| mix_weights: Vector::ones(k) / (k <span class="kw">as </span>f64), |
| model_means: <span class="prelude-val">None</span>, |
| model_covars: <span class="prelude-val">None</span>, |
| log_lik: <span class="number">0f64</span>, |
| max_iters: <span class="number">100</span>, |
| cov_option: CovOption::Full, |
| } |
| } |
| |
| <span class="doccomment">/// Constructs a new GMM with the specified prior mixture weights. |
| /// |
| /// The mixture weights must have the same length as the number of components. |
| /// Each element of the mixture weights must be non-negative. |
| /// |
| /// # Examples |
| /// |
| /// ``` |
| /// use rusty_machine::learning::gmm::GaussianMixtureModel; |
| /// use rusty_machine::linalg::Vector; |
| /// |
| /// let mix_weights = Vector::new(vec![0.25, 0.25, 0.5]); |
| /// |
| /// let gmm = GaussianMixtureModel::with_weights(3, mix_weights).unwrap(); |
| /// ``` |
| /// |
| /// # Failures |
| /// |
| /// Fails if either of the following conditions are met: |
| /// |
| /// - Mixture weights do not have length k. |
| /// - Mixture weights have a negative entry. |
| </span><span class="kw">pub fn </span>with_weights(k: usize, mixture_weights: Vector<f64>) -> LearningResult<GaussianMixtureModel> { |
| <span class="kw">if </span>mixture_weights.size() != k { |
| <span class="prelude-val">Err</span>(Error::new(ErrorKind::InvalidParameters, <span class="string">"Mixture weights must have length k."</span>)) |
| } <span class="kw">else if </span>mixture_weights.data().iter().any(|<span class="kw-2">&</span>x| x < <span class="number">0f64</span>) { |
| <span class="prelude-val">Err</span>(Error::new(ErrorKind::InvalidParameters, <span class="string">"Mixture weights must have only non-negative entries."</span>)) |
| } <span class="kw">else </span>{ |
| <span class="kw">let </span>sum = mixture_weights.sum(); |
| <span class="kw">let </span>normalized_weights = mixture_weights / sum; |
| |
| <span class="prelude-val">Ok</span>(GaussianMixtureModel { |
| comp_count: k, |
| mix_weights: normalized_weights, |
| model_means: <span class="prelude-val">None</span>, |
| model_covars: <span class="prelude-val">None</span>, |
| log_lik: <span class="number">0f64</span>, |
| max_iters: <span class="number">100</span>, |
| cov_option: CovOption::Full, |
| }) |
| } |
| } |
| |
| <span class="doccomment">/// The model means |
| /// |
| /// Returns an Option<&Matrix<f64>> containing |
| /// the model means. Each row represents |
| /// the mean of one of the Gaussians. |
| </span><span class="kw">pub fn </span>means(<span class="kw-2">&</span><span class="self">self</span>) -> <span class="prelude-ty">Option</span><<span class="kw-2">&</span>Matrix<f64>> { |
| <span class="self">self</span>.model_means.as_ref() |
| } |
| |
| <span class="doccomment">/// The model covariances |
| /// |
| /// Returns an Option<&Vec<Matrix<f64>>> containing |
| /// the model covariances. Each Matrix in the vector |
| /// is the covariance of one of the Gaussians. |
| </span><span class="kw">pub fn </span>covariances(<span class="kw-2">&</span><span class="self">self</span>) -> <span class="prelude-ty">Option</span><<span class="kw-2">&</span>Vec<Matrix<f64>>> { |
| <span class="self">self</span>.model_covars.as_ref() |
| } |
| |
| <span class="doccomment">/// The model mixture weights |
| /// |
| /// Returns a reference to the model mixture weights. |
| /// These are the weighted contributions of each underlying |
| /// Gaussian to the model distribution. |
| </span><span class="kw">pub fn </span>mixture_weights(<span class="kw-2">&</span><span class="self">self</span>) -> <span class="kw-2">&</span>Vector<f64> { |
| <span class="kw-2">&</span><span class="self">self</span>.mix_weights |
| } |
| |
| <span class="doccomment">/// Sets the max number of iterations for the EM algorithm. |
| /// |
| /// # Examples |
| /// |
| /// ``` |
| /// use rusty_machine::learning::gmm::GaussianMixtureModel; |
| /// |
| /// let mut gmm = GaussianMixtureModel::new(2); |
| /// gmm.set_max_iters(5); |
| /// ``` |
| </span><span class="kw">pub fn </span>set_max_iters(<span class="kw-2">&mut </span><span class="self">self</span>, iters: usize) { |
| <span class="self">self</span>.max_iters = iters; |
| } |
| |
| <span class="kw">fn </span>initialize_covariances(<span class="kw-2">&</span><span class="self">self</span>, inputs: <span class="kw-2">&</span>Matrix<f64>, reg_value: f64) -> LearningResult<Matrix<f64>> { |
| <span class="kw">match </span><span class="self">self</span>.cov_option { |
| CovOption::Diagonal => { |
| <span class="kw">let </span>variance = inputs.variance(Axes::Row)<span class="question-mark">?</span>; |
| <span class="prelude-val">Ok</span>(Matrix::from_diag(variance.data()) * reg_value.sqrt()) |
| } |
| |
| CovOption::Full | CovOption::Regularized(<span class="kw">_</span>) => { |
| <span class="kw">let </span>means = inputs.mean(Axes::Row); |
| <span class="kw">let </span><span class="kw-2">mut </span>cov_mat = Matrix::zeros(inputs.cols(), inputs.cols()); |
| <span class="kw">for </span>(j, <span class="kw-2">mut </span>row) <span class="kw">in </span>cov_mat.row_iter_mut().enumerate() { |
| <span class="kw">for </span>(k, elem) <span class="kw">in </span>row.iter_mut().enumerate() { |
| <span class="kw-2">*</span>elem = inputs.row_iter().map(|r| { |
| (r[j] - means[j]) * (r[k] - means[k]) |
| }).sum::<f64>(); |
| } |
| } |
| cov_mat <span class="kw-2">*</span>= reg_value; |
| <span class="kw">if let </span>CovOption::Regularized(eps) = <span class="self">self</span>.cov_option { |
| cov_mat += Matrix::<f64>::identity(cov_mat.cols()) * eps; |
| } |
| <span class="prelude-val">Ok</span>(cov_mat) |
| } |
| } |
| } |
| |
| <span class="kw">fn </span>membership_weights(<span class="kw-2">&</span><span class="self">self</span>, inputs: <span class="kw-2">&</span>Matrix<f64>) -> LearningResult<(Matrix<f64>, f64)> { |
| <span class="kw">let </span>n = inputs.rows(); |
| |
| <span class="kw">let </span><span class="kw-2">mut </span>member_weights_data = Vec::with_capacity(n * <span class="self">self</span>.comp_count); |
| |
| <span class="comment">// We compute the determinants and inverses now |
| </span><span class="kw">let </span><span class="kw-2">mut </span>cov_sqrt_dets = Vec::with_capacity(<span class="self">self</span>.comp_count); |
| <span class="kw">let </span><span class="kw-2">mut </span>cov_invs = Vec::with_capacity(<span class="self">self</span>.comp_count); |
| |
| <span class="kw">if let </span><span class="prelude-val">Some</span>(<span class="kw-2">ref </span>covars) = <span class="self">self</span>.model_covars { |
| <span class="kw">for </span>cov <span class="kw">in </span>covars { |
| <span class="kw">let </span>lup = PartialPivLu::decompose(cov.clone()).expect(<span class="string">"Covariance could not be lup decomposed"</span>); |
| <span class="kw">let </span>covar_det = lup.det(); |
| <span class="comment">// TODO: We can probably remove this inverse for a more stable solve elsewhere. |
| </span><span class="kw">let </span>covar_inv = lup.inverse().map_err(Error::from)<span class="question-mark">?</span>; |
| |
| cov_sqrt_dets.push(covar_det.sqrt()); |
| cov_invs.push(covar_inv); |
| } |
| } |
| |
| <span class="kw">let </span><span class="kw-2">mut </span>log_lik = <span class="number">0f64</span>; |
| |
| <span class="comment">// Now we compute the membership weights |
| </span><span class="kw">if let </span><span class="prelude-val">Some</span>(<span class="kw-2">ref </span>means) = <span class="self">self</span>.model_means { |
| <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..n { |
| <span class="kw">let </span><span class="kw-2">mut </span>pdfs = Vec::with_capacity(<span class="self">self</span>.comp_count); |
| <span class="kw">let </span>x_i = MatrixSlice::from_matrix(inputs, [i, <span class="number">0</span>], <span class="number">1</span>, inputs.cols()); |
| |
| <span class="kw">for </span>j <span class="kw">in </span><span class="number">0</span>..<span class="self">self</span>.comp_count { |
| <span class="kw">let </span>mu_j = MatrixSlice::from_matrix(means, [j, <span class="number">0</span>], <span class="number">1</span>, means.cols()); |
| <span class="kw">let </span>diff = x_i - mu_j; |
| |
| <span class="kw">let </span>pdf = (<span class="kw-2">&</span>diff * <span class="kw-2">&</span>cov_invs[j] * diff.transpose() * -<span class="number">0.5</span>).into_vec()[<span class="number">0</span>] |
| .exp() / cov_sqrt_dets[j]; |
| pdfs.push(pdf); |
| } |
| |
| <span class="kw">let </span>weighted_pdf_sum = utils::dot(<span class="kw-2">&</span>pdfs, <span class="self">self</span>.mix_weights.data()); |
| |
| <span class="kw">for </span>(idx, pdf) <span class="kw">in </span>pdfs.iter().enumerate() { |
| member_weights_data.push(<span class="self">self</span>.mix_weights[idx] * pdf / (weighted_pdf_sum)); |
| } |
| |
| log_lik += weighted_pdf_sum.ln(); |
| } |
| } |
| |
| <span class="prelude-val">Ok</span>((Matrix::new(n, <span class="self">self</span>.comp_count, member_weights_data), log_lik)) |
| } |
| |
| <span class="kw">fn </span>update_params(<span class="kw-2">&mut </span><span class="self">self</span>, inputs: <span class="kw-2">&</span>Matrix<f64>, membership_weights: Matrix<f64>) { |
| <span class="kw">let </span>n = membership_weights.rows(); |
| <span class="kw">let </span>d = inputs.cols(); |
| |
| <span class="kw">let </span>sum_weights = membership_weights.sum_rows(); |
| |
| <span class="self">self</span>.mix_weights = <span class="kw-2">&</span>sum_weights / (n <span class="kw">as </span>f64); |
| |
| <span class="kw">let </span><span class="kw-2">mut </span>new_means = membership_weights.transpose() * inputs; |
| |
| <span class="kw">for </span>(<span class="kw-2">mut </span>mean, w) <span class="kw">in </span>new_means.row_iter_mut().zip(sum_weights.data().iter()) { |
| <span class="kw-2">*</span>mean /= <span class="kw-2">*</span>w; |
| } |
| |
| <span class="kw">let </span><span class="kw-2">mut </span>new_covs = Vec::with_capacity(<span class="self">self</span>.comp_count); |
| |
| <span class="kw">for </span>k <span class="kw">in </span><span class="number">0</span>..<span class="self">self</span>.comp_count { |
| <span class="kw">let </span><span class="kw-2">mut </span>cov_mat = Matrix::zeros(d, d); |
| <span class="kw">let </span>new_means_k = MatrixSlice::from_matrix(<span class="kw-2">&</span>new_means, [k, <span class="number">0</span>], <span class="number">1</span>, d); |
| |
| <span class="kw">for </span>i <span class="kw">in </span><span class="number">0</span>..n { |
| <span class="kw">let </span>inputs_i = MatrixSlice::from_matrix(inputs, [i, <span class="number">0</span>], <span class="number">1</span>, d); |
| <span class="kw">let </span>diff = inputs_i - new_means_k; |
| cov_mat += <span class="self">self</span>.compute_cov(diff, membership_weights[[i, k]]); |
| } |
| |
| <span class="kw">if let </span>CovOption::Regularized(eps) = <span class="self">self</span>.cov_option { |
| cov_mat += Matrix::<f64>::identity(cov_mat.cols()) * eps; |
| } |
| |
| new_covs.push(cov_mat / sum_weights[k]); |
| |
| } |
| |
| <span class="self">self</span>.model_means = <span class="prelude-val">Some</span>(new_means); |
| <span class="self">self</span>.model_covars = <span class="prelude-val">Some</span>(new_covs); |
| } |
| |
| <span class="kw">fn </span>compute_cov(<span class="kw-2">&</span><span class="self">self</span>, diff: Matrix<f64>, weight: f64) -> Matrix<f64> { |
| <span class="kw">match </span><span class="self">self</span>.cov_option { |
| CovOption::Full | CovOption::Regularized(<span class="kw">_</span>) => (diff.transpose() * diff) * weight, |
| CovOption::Diagonal => Matrix::from_diag(<span class="kw-2">&</span>diff.elemul(<span class="kw-2">&</span>diff).into_vec()) * weight, |
| } |
| } |
| } |
| |
| <span class="attribute">#[cfg(test)] |
| </span><span class="kw">mod </span>tests { |
| <span class="kw">use </span><span class="kw">super</span>::GaussianMixtureModel; |
| <span class="kw">use </span>linalg::Vector; |
| |
| <span class="attribute">#[test] |
| </span><span class="kw">fn </span>test_means_none() { |
| <span class="kw">let </span>model = GaussianMixtureModel::new(<span class="number">5</span>); |
| |
| <span class="macro">assert_eq!</span>(model.means(), <span class="prelude-val">None</span>); |
| } |
| |
| <span class="attribute">#[test] |
| </span><span class="kw">fn </span>test_covars_none() { |
| <span class="kw">let </span>model = GaussianMixtureModel::new(<span class="number">5</span>); |
| |
| <span class="macro">assert_eq!</span>(model.covariances(), <span class="prelude-val">None</span>); |
| } |
| |
| <span class="attribute">#[test] |
| </span><span class="kw">fn </span>test_negative_mixtures() { |
| <span class="kw">let </span>mix_weights = Vector::new(<span class="macro">vec!</span>[-<span class="number">0.25</span>, <span class="number">0.75</span>, <span class="number">0.5</span>]); |
| <span class="kw">let </span>gmm_res = GaussianMixtureModel::with_weights(<span class="number">3</span>, mix_weights); |
| <span class="macro">assert!</span>(gmm_res.is_err()); |
| } |
| |
| <span class="attribute">#[test] |
| </span><span class="kw">fn </span>test_wrong_length_mixtures() { |
| <span class="kw">let </span>mix_weights = Vector::new(<span class="macro">vec!</span>[<span class="number">0.1</span>, <span class="number">0.25</span>, <span class="number">0.75</span>, <span class="number">0.5</span>]); |
| <span class="kw">let </span>gmm_res = GaussianMixtureModel::with_weights(<span class="number">3</span>, mix_weights); |
| <span class="macro">assert!</span>(gmm_res.is_err()); |
| } |
| } |
| </code></pre></div> |
| </section></div></main><div id="rustdoc-vars" data-root-path="../../../" data-current-crate="rusty_machine" data-themes="ayu,dark,light" data-resource-suffix="" data-rustdoc-version="1.66.0-nightly (5c8bff74b 2022-10-21)" ></div></body></html> |