|
a |
|
b/docs/articles/using_tb.html |
|
|
1 |
<!DOCTYPE html> |
|
|
2 |
<!-- Generated by pkgdown: do not edit by hand --><html lang="en"> |
|
|
3 |
<head> |
|
|
4 |
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> |
|
|
5 |
<meta charset="utf-8"> |
|
|
6 |
<meta http-equiv="X-UA-Compatible" content="IE=edge"> |
|
|
7 |
<meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"> |
|
|
8 |
<meta name="description" content="deepG"> |
|
|
9 |
<title>Using Tensorboard • deepG</title> |
|
|
10 |
<!-- favicons --><link rel="icon" type="image/png" sizes="16x16" href="../favicon-16x16.png"> |
|
|
11 |
<link rel="icon" type="image/png" sizes="32x32" href="../favicon-32x32.png"> |
|
|
12 |
<link rel="apple-touch-icon" type="image/png" sizes="180x180" href="../apple-touch-icon.png"> |
|
|
13 |
<link rel="apple-touch-icon" type="image/png" sizes="120x120" href="../apple-touch-icon-120x120.png"> |
|
|
14 |
<link rel="apple-touch-icon" type="image/png" sizes="76x76" href="../apple-touch-icon-76x76.png"> |
|
|
15 |
<link rel="apple-touch-icon" type="image/png" sizes="60x60" href="../apple-touch-icon-60x60.png"> |
|
|
16 |
<script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"> |
|
|
17 |
<link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"> |
|
|
18 |
<script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><!-- Font Awesome icons --><link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/5.12.1/css/all.min.css" integrity="sha256-mmgLkCYLUQbXn0B1SRqzHar6dCnv9oZFPEC1g1cwlkk=" crossorigin="anonymous"> |
|
|
19 |
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/5.12.1/css/v4-shims.min.css" integrity="sha256-wZjR52fzng1pJHwx4aV2AO3yyTOXrcDW7jBpJtTwVxw=" crossorigin="anonymous"> |
|
|
20 |
<!-- bootstrap-toc --><script src="https://cdn.jsdelivr.net/gh/afeld/bootstrap-toc@v1.0.1/dist/bootstrap-toc.min.js" integrity="sha256-4veVQbu7//Lk5TSmc7YV48MxtMy98e26cf5MrgZYnwo=" crossorigin="anonymous"></script><!-- headroom.js --><script src="https://cdnjs.cloudflare.com/ajax/libs/headroom/0.11.0/headroom.min.js" integrity="sha256-AsUX4SJE1+yuDu5+mAVzJbuYNPHj/WroHuZ8Ir/CkE0=" crossorigin="anonymous"></script><script src="https://cdnjs.cloudflare.com/ajax/libs/headroom/0.11.0/jQuery.headroom.min.js" integrity="sha256-ZX/yNShbjqsohH1k95liqY9Gd8uOiE1S4vZc+9KQ1K4=" crossorigin="anonymous"></script><!-- clipboard.js --><script src="https://cdnjs.cloudflare.com/ajax/libs/clipboard.js/2.0.11/clipboard.min.js" integrity="sha512-7O5pXpc0oCRrxk8RUfDYFgn0nO1t+jLuIOQdOMRp4APB7uZ4vSjspzp5y6YDtDs4VzUSTbWzBFZ/LKJhnyFOKw==" crossorigin="anonymous" referrerpolicy="no-referrer"></script><!-- search --><script src="https://cdnjs.cloudflare.com/ajax/libs/fuse.js/6.4.6/fuse.js" integrity="sha512-zv6Ywkjyktsohkbp9bb45V6tEMoWhzFzXis+LrMehmJZZSys19Yxf1dopHx7WzIKxr5tK2dVcYmaCk2uqdjF4A==" crossorigin="anonymous"></script><script src="https://cdnjs.cloudflare.com/ajax/libs/autocomplete.js/0.38.0/autocomplete.jquery.min.js" integrity="sha512-GU9ayf+66Xx2TmpxqJpliWbT5PiGYxpaG8rfnBEk1LL8l1KGkRShhngwdXK1UgqhAzWpZHSiYPc09/NwDQIGyg==" crossorigin="anonymous"></script><script src="https://cdnjs.cloudflare.com/ajax/libs/mark.js/8.11.1/mark.min.js" integrity="sha512-5CYOlHXGh6QpOFA/TeTylKLWfB3ftPsde7AnmhuitiTX4K5SqCLBeKro6sPS8ilsz1Q4NRx3v8Ko2IBiszzdww==" crossorigin="anonymous"></script><!-- pkgdown --><script src="../pkgdown.js"></script><meta property="og:title" content="Using Tensorboard"> |
|
|
21 |
<meta property="og:description" content="deepG"> |
|
|
22 |
<meta property="og:image" content="https://genomenet.github.io/deepG/logo.png"> |
|
|
23 |
<!-- mathjax --><script src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js" integrity="sha256-nvJJv9wWKEm88qvoQl9ekL2J+k/RWIsaSScxxlsrv8k=" crossorigin="anonymous"></script><script src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/config/TeX-AMS-MML_HTMLorMML.js" integrity="sha256-84DKXVJXs0/F8OTMzX4UR909+jtl4G7SPypPavF+GfA=" crossorigin="anonymous"></script><!--[if lt IE 9]> |
|
|
24 |
<script src="https://oss.maxcdn.com/html5shiv/3.7.3/html5shiv.min.js"></script> |
|
|
25 |
<script src="https://oss.maxcdn.com/respond/1.4.2/respond.min.js"></script> |
|
|
26 |
<![endif]--> |
|
|
27 |
</head> |
|
|
28 |
<body> |
|
|
29 |
<a href="#main" class="visually-hidden-focusable">Skip to contents</a> |
|
|
30 |
|
|
|
31 |
|
|
|
32 |
<nav class="navbar fixed-top navbar-light navbar-expand-lg bg-light" data-bs-theme="light"><div class="container"> |
|
|
33 |
|
|
|
34 |
<a class="navbar-brand me-2" href="../index.html">deepG</a> |
|
|
35 |
|
|
|
36 |
<small class="nav-text text-default me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="Released version">0.3.0</small> |
|
|
37 |
|
|
|
38 |
|
|
|
39 |
<button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation"> |
|
|
40 |
<span class="navbar-toggler-icon"></span> |
|
|
41 |
</button> |
|
|
42 |
|
|
|
43 |
<div id="navbar" class="collapse navbar-collapse ms-3"> |
|
|
44 |
<ul class="navbar-nav me-auto"> |
|
|
45 |
<li class="nav-item"> |
|
|
46 |
<a class="nav-link" href="../reference/index.html"> |
|
|
47 |
<span class="fa fa fa fa-file-alt"></span> |
|
|
48 |
|
|
|
49 |
Reference |
|
|
50 |
</a> |
|
|
51 |
</li> |
|
|
52 |
<li class="nav-item dropdown"> |
|
|
53 |
<a href="#" class="nav-link dropdown-toggle" data-bs-toggle="dropdown" role="button" aria-expanded="false" aria-haspopup="true" id="dropdown-notebooks">Notebooks</a> |
|
|
54 |
<div class="dropdown-menu" aria-labelledby="dropdown-notebooks"> |
|
|
55 |
<a class="external-link dropdown-item" href="https://colab.research.google.com/drive/175jIdXcDcgPUvaBo2rH2Lupbpjnp5O7G?usp=sharing">deepG tutorial</a> |
|
|
56 |
<a class="external-link dropdown-item" href="https://colab.research.google.com/drive/1Eolc0koMNM1zkuO4XyVM58ImeF1BpRiH?usp=sharing">Read-length level: Human contamination</a> |
|
|
57 |
<a class="external-link dropdown-item" href="https://colab.research.google.com/drive/1yiXSwFafXpMLHaov9iBTQLIDZ6bK1zYX?usp=sharing">Locus level: CRISPR detection</a> |
|
|
58 |
<a class="external-link dropdown-item" href="https://colab.research.google.com/drive/1G7bOFEX87cZNrM2tdRtTdkrZn5fM__g0?usp=sharing">Gene level: 16S rRNA detection</a> |
|
|
59 |
<a class="external-link dropdown-item" href="https://colab.research.google.com/drive/1BCggL-tfQF136YeJ8cKKi-zoBEDMgkNh?usp=sharing">Genome level: Bacterial morphology (Sporulation)</a> |
|
|
60 |
<a class="external-link dropdown-item" href="https://colab.research.google.com/drive/10xpRzGd3JeBAbqQYSCxzQUMctt01sx9D?usp=sharing">Full metagenome level: Colorectal cancer prediction</a> |
|
|
61 |
<a class="external-link dropdown-item" href="https://colab.research.google.com/drive/1kyYK7IU7GSfdpDzO_a8U3_qD4i3zTu6w?usp=sharing">BERT with deepG</a> |
|
|
62 |
</div> |
|
|
63 |
</li> |
|
|
64 |
<li class="active nav-item dropdown"> |
|
|
65 |
<a href="#" class="nav-link dropdown-toggle" data-bs-toggle="dropdown" role="button" aria-expanded="false" aria-haspopup="true" id="dropdown-tutorials">Tutorials</a> |
|
|
66 |
<div class="dropdown-menu" aria-labelledby="dropdown-tutorials"> |
|
|
67 |
<a class="dropdown-item" href="../articles/getting_started.html">Getting Started</a> |
|
|
68 |
<a class="dropdown-item" href="../articles/training_types.html">Training types</a> |
|
|
69 |
<a class="dropdown-item" href="../articles/data_generator.html">Data generator</a> |
|
|
70 |
<a class="dropdown-item" href="../articles/using_tb.html">Using tensorboard</a> |
|
|
71 |
<a class="dropdown-item" href="../articles/integrated_gradient.html">Integrated Gradient</a> |
|
|
72 |
</div> |
|
|
73 |
</li> |
|
|
74 |
</ul> |
|
|
75 |
<form class="form-inline my-2 my-lg-0" role="search"> |
|
|
76 |
<input type="search" class="form-control me-sm-2" aria-label="Toggle navigation" name="search-input" data-search-index="../search.json" id="search-input" placeholder="Search for" autocomplete="off"> |
|
|
77 |
</form> |
|
|
78 |
|
|
|
79 |
<ul class="navbar-nav"> |
|
|
80 |
<li class="nav-item"> |
|
|
81 |
<a class="external-link nav-link" href="https://github.com/GenomeNet/deepG/" aria-label="github"> |
|
|
82 |
<span class="fab fa fab fa-github fa-lg"></span> |
|
|
83 |
|
|
|
84 |
</a> |
|
|
85 |
</li> |
|
|
86 |
</ul> |
|
|
87 |
</div> |
|
|
88 |
|
|
|
89 |
|
|
|
90 |
</div> |
|
|
91 |
</nav><div class="container template-article"> |
|
|
92 |
|
|
|
93 |
|
|
|
94 |
|
|
|
95 |
|
|
|
96 |
<div class="row"> |
|
|
97 |
<main id="main" class="col-md-9"><div class="page-header"> |
|
|
98 |
<img src="../logo.png" class="logo" alt=""><h1>Using Tensorboard</h1> |
|
|
99 |
|
|
|
100 |
|
|
|
101 |
<small class="dont-index">Source: <a href="https://github.com/GenomeNet/deepG/blob/HEAD/vignettes/using_tb.Rmd" class="external-link"><code>vignettes/using_tb.Rmd</code></a></small> |
|
|
102 |
<div class="d-none name"><code>using_tb.Rmd</code></div> |
|
|
103 |
</div> |
|
|
104 |
|
|
|
105 |
|
|
|
106 |
|
|
|
107 |
<div class="sourceCode" id="cb1"><pre class="downlit sourceCode r"> |
|
|
108 |
<code class="sourceCode R"><span><span class="co"># devtools::install_github("GenomeNet/deepG")</span></span> |
|
|
109 |
<span><span class="kw"><a href="https://rdrr.io/r/base/library.html" class="external-link">library</a></span><span class="op">(</span><span class="va"><a href="https://github.com/GenomeNet/deepG" class="external-link">deepG</a></span><span class="op">)</span></span> |
|
|
110 |
<span><span class="kw"><a href="https://rdrr.io/r/base/library.html" class="external-link">library</a></span><span class="op">(</span><span class="va"><a href="https://magrittr.tidyverse.org" class="external-link">magrittr</a></span><span class="op">)</span></span></code></pre></div> |
|
|
111 |
<style type="text/css"> |
|
|
112 |
mark.in { |
|
|
113 |
background-color: CornflowerBlue; |
|
|
114 |
} |
|
|
115 |
|
|
|
116 |
mark.out { |
|
|
117 |
background-color: IndianRed; |
|
|
118 |
} |
|
|
119 |
|
|
|
120 |
</style> |
|
|
121 |
<p>Tensorflow offers the |
|
|
122 |
<a href="https://www.tensorflow.org/tensorboard" class="external-link">Tensorboard</a> |
|
|
123 |
application to visualize the training process of our networks. DeepG |
|
|
124 |
expands on some of the default Tensorboard options and implements some |
|
|
125 |
custom settings. We train again a model that can differentiate sequences |
|
|
126 |
based on the GC content, as described in the |
|
|
127 |
<a href="getting_started.html">Getting started tutorial</a>.</p> |
|
|
128 |
<p>We start by creating our data. To show the difference between |
|
|
129 |
accuracy and balanced accuracy, we create 3 times more data with high GC |
|
|
130 |
content.</p> |
|
|
131 |
<div class="sourceCode" id="cb2"><pre class="downlit sourceCode r"> |
|
|
132 |
<code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/Random.html" class="external-link">set.seed</a></span><span class="op">(</span><span class="fl">123</span><span class="op">)</span></span> |
|
|
133 |
<span><span class="va">vocabulary</span> <span class="op"><-</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="st">"A"</span>, <span class="st">"C"</span>, <span class="st">"G"</span>, <span class="st">"T"</span><span class="op">)</span></span> |
|
|
134 |
<span></span> |
|
|
135 |
<span><span class="va">data_type</span> <span class="op"><-</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="st">"train"</span>, <span class="st">"validation"</span><span class="op">)</span></span> |
|
|
136 |
<span></span> |
|
|
137 |
<span><span class="kw">for</span> <span class="op">(</span><span class="va">i</span> <span class="kw">in</span> <span class="fl">1</span><span class="op">:</span><span class="fu"><a href="https://rdrr.io/r/base/length.html" class="external-link">length</a></span><span class="op">(</span><span class="va">data_type</span><span class="op">)</span><span class="op">)</span> <span class="op">{</span></span> |
|
|
138 |
<span> </span> |
|
|
139 |
<span> <span class="va">temp_file</span> <span class="op"><-</span> <span class="fu"><a href="https://rdrr.io/r/base/tempfile.html" class="external-link">tempfile</a></span><span class="op">(</span><span class="op">)</span></span> |
|
|
140 |
<span> <span class="fu"><a href="https://rdrr.io/r/base/assign.html" class="external-link">assign</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/paste.html" class="external-link">paste0</a></span><span class="op">(</span><span class="va">data_type</span><span class="op">[</span><span class="va">i</span><span class="op">]</span>, <span class="st">"_dir"</span><span class="op">)</span>, <span class="va">temp_file</span><span class="op">)</span></span> |
|
|
141 |
<span> <span class="fu"><a href="https://rdrr.io/r/base/files2.html" class="external-link">dir.create</a></span><span class="op">(</span><span class="va">temp_file</span><span class="op">)</span></span> |
|
|
142 |
<span> </span> |
|
|
143 |
<span> <span class="kw">for</span> <span class="op">(</span><span class="va">j</span> <span class="kw">in</span> <span class="fl">1</span><span class="op">:</span><span class="fl">6</span><span class="op">)</span> <span class="op">{</span></span> |
|
|
144 |
<span> </span> |
|
|
145 |
<span> <span class="kw">if</span> <span class="op">(</span><span class="va">j</span> <span class="op"><a href="https://rdrr.io/r/base/Arithmetic.html" class="external-link">%%</a></span> <span class="fl">2</span> <span class="op">==</span> <span class="fl">1</span><span class="op">)</span> <span class="op">{</span></span> |
|
|
146 |
<span> <span class="va">header</span> <span class="op"><-</span> <span class="st">"high_gc"</span></span> |
|
|
147 |
<span> <span class="va">prob</span> <span class="op"><-</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.1</span>, <span class="fl">0.4</span>, <span class="fl">0.4</span>, <span class="fl">0.1</span><span class="op">)</span></span> |
|
|
148 |
<span> <span class="op">}</span> <span class="kw">else</span> <span class="op">{</span></span> |
|
|
149 |
<span> <span class="va">header</span> <span class="op"><-</span> <span class="st">"equal_dist"</span></span> |
|
|
150 |
<span> <span class="va">prob</span> <span class="op"><-</span> <span class="fu"><a href="https://rdrr.io/r/base/rep.html" class="external-link">rep</a></span><span class="op">(</span><span class="fl">0.25</span>, <span class="fl">4</span><span class="op">)</span></span> |
|
|
151 |
<span> <span class="op">}</span></span> |
|
|
152 |
<span> <span class="va">fasta_name_start</span> <span class="op"><-</span> <span class="fu"><a href="https://rdrr.io/r/base/paste.html" class="external-link">paste0</a></span><span class="op">(</span><span class="va">header</span>, <span class="st">"_"</span>, <span class="va">data_type</span><span class="op">[</span><span class="va">i</span><span class="op">]</span><span class="op">)</span></span> |
|
|
153 |
<span> </span> |
|
|
154 |
<span> <span class="fu"><a href="../reference/create_dummy_data.html">create_dummy_data</a></span><span class="op">(</span>file_path <span class="op">=</span> <span class="va">temp_file</span>,</span> |
|
|
155 |
<span> num_files <span class="op">=</span> <span class="fl">2</span>,</span> |
|
|
156 |
<span> seq_length <span class="op">=</span> <span class="fl">100</span>, </span> |
|
|
157 |
<span> num_seq <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/ifelse.html" class="external-link">ifelse</a></span><span class="op">(</span><span class="va">j</span> <span class="op"><a href="https://rdrr.io/r/base/Arithmetic.html" class="external-link">%%</a></span> <span class="fl">2</span> <span class="op">==</span> <span class="fl">1</span>, <span class="fl">6</span>, <span class="fl">2</span><span class="op">)</span>, <span class="co"># create more sequences for high GC content</span></span> |
|
|
158 |
<span> header <span class="op">=</span> <span class="va">header</span>,</span> |
|
|
159 |
<span> prob <span class="op">=</span> <span class="va">prob</span>,</span> |
|
|
160 |
<span> fasta_name_start <span class="op">=</span> <span class="va">fasta_name_start</span>,</span> |
|
|
161 |
<span> vocabulary <span class="op">=</span> <span class="va">vocabulary</span><span class="op">)</span></span> |
|
|
162 |
<span> <span class="op">}</span></span> |
|
|
163 |
<span> </span> |
|
|
164 |
<span><span class="op">}</span></span></code></pre></div> |
|
|
165 |
<div class="sourceCode" id="cb3"><pre class="downlit sourceCode r"> |
|
|
166 |
<code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/list.files.html" class="external-link">list.files</a></span><span class="op">(</span><span class="va">train_dir</span><span class="op">)</span></span></code></pre></div> |
|
|
167 |
<pre><code><span><span class="co">## [1] "equal_dist_train_1.fasta" "equal_dist_train_2.fasta"</span></span> |
|
|
168 |
<span><span class="co">## [3] "high_gc_train_1.fasta" "high_gc_train_2.fasta"</span></span></code></pre> |
|
|
169 |
<div class="sourceCode" id="cb5"><pre class="downlit sourceCode r"> |
|
|
170 |
<code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/list.files.html" class="external-link">list.files</a></span><span class="op">(</span><span class="va">validation_dir</span><span class="op">)</span></span></code></pre></div> |
|
|
171 |
<pre><code><span><span class="co">## [1] "equal_dist_validation_1.fasta" "equal_dist_validation_2.fasta"</span></span> |
|
|
172 |
<span><span class="co">## [3] "high_gc_validation_1.fasta" "high_gc_validation_2.fasta"</span></span></code></pre> |
|
|
173 |
<p>To use tensorboard, we first need to create a folder where we store |
|
|
174 |
our tensorboard logs.</p> |
|
|
175 |
<div class="sourceCode" id="cb7"><pre class="downlit sourceCode r"> |
|
|
176 |
<code class="sourceCode R"><span><span class="co"># create folder for tensorboard logs</span></span> |
|
|
177 |
<span><span class="va">tb_dir</span> <span class="op"><-</span> <span class="fu"><a href="https://rdrr.io/r/base/tempfile.html" class="external-link">tempfile</a></span><span class="op">(</span><span class="op">)</span></span> |
|
|
178 |
<span><span class="fu"><a href="https://rdrr.io/r/base/files2.html" class="external-link">dir.create</a></span><span class="op">(</span><span class="va">tb_dir</span><span class="op">)</span></span></code></pre></div> |
|
|
179 |
<p>When creating our model, we can add some additional metrics to |
|
|
180 |
observe like AUC, F1 and balanced accuracy.</p> |
|
|
181 |
<div class="sourceCode" id="cb8"><pre class="downlit sourceCode r"> |
|
|
182 |
<code class="sourceCode R"><span><span class="va">maxlen</span> <span class="op"><-</span> <span class="fl">50</span></span> |
|
|
183 |
<span><span class="va">model</span> <span class="op"><-</span> <span class="fu"><a href="../reference/create_model_lstm_cnn.html">create_model_lstm_cnn</a></span><span class="op">(</span>maxlen <span class="op">=</span> <span class="va">maxlen</span>,</span> |
|
|
184 |
<span> filters <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">8</span><span class="op">)</span>,</span> |
|
|
185 |
<span> kernel_size <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">12</span><span class="op">)</span>,</span> |
|
|
186 |
<span> pool_size <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">3</span><span class="op">)</span>,</span> |
|
|
187 |
<span> layer_lstm <span class="op">=</span> <span class="fl">8</span>,</span> |
|
|
188 |
<span> auc_metric <span class="op">=</span> <span class="cn">TRUE</span>,</span> |
|
|
189 |
<span> f1_metric <span class="op">=</span> <span class="cn">TRUE</span>,</span> |
|
|
190 |
<span> bal_acc <span class="op">=</span> <span class="cn">TRUE</span>,</span> |
|
|
191 |
<span> layer_dense <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">4</span>, <span class="fl">2</span><span class="op">)</span>,</span> |
|
|
192 |
<span> model_seed <span class="op">=</span> <span class="fl">3</span><span class="op">)</span></span></code></pre></div> |
|
|
193 |
<pre><code><span><span class="co">## Will flatten y_true and y_pred matrices to a single vector for evaluation</span></span> |
|
|
194 |
<span><span class="co">## rather than computing separate F1 scores for each class and taking the mean.</span></span></code></pre> |
|
|
195 |
<pre><code><span><span class="co">## Model: "model"</span></span> |
|
|
196 |
<span><span class="co">## _________________________________________________________________</span></span> |
|
|
197 |
<span><span class="co">## Layer (type) Output Shape Param # </span></span> |
|
|
198 |
<span><span class="co">## =================================================================</span></span> |
|
|
199 |
<span><span class="co">## input_1 (InputLayer) [(None, 50, 4)] 0 </span></span> |
|
|
200 |
<span><span class="co">## </span></span> |
|
|
201 |
<span><span class="co">## conv1d (Conv1D) (None, 50, 8) 392 </span></span> |
|
|
202 |
<span><span class="co">## </span></span> |
|
|
203 |
<span><span class="co">## max_pooling1d (MaxPooling1 (None, 16, 8) 0 </span></span> |
|
|
204 |
<span><span class="co">## D) </span></span> |
|
|
205 |
<span><span class="co">## </span></span> |
|
|
206 |
<span><span class="co">## batch_normalization (Batch (None, 16, 8) 32 </span></span> |
|
|
207 |
<span><span class="co">## Normalization) </span></span> |
|
|
208 |
<span><span class="co">## </span></span> |
|
|
209 |
<span><span class="co">## lstm (LSTM) (None, 8) 544 </span></span> |
|
|
210 |
<span><span class="co">## </span></span> |
|
|
211 |
<span><span class="co">## dense (Dense) (None, 4) 36 </span></span> |
|
|
212 |
<span><span class="co">## </span></span> |
|
|
213 |
<span><span class="co">## dense_1 (Dense) (None, 2) 10 </span></span> |
|
|
214 |
<span><span class="co">## </span></span> |
|
|
215 |
<span><span class="co">## =================================================================</span></span> |
|
|
216 |
<span><span class="co">## Total params: 1014 (3.96 KB)</span></span> |
|
|
217 |
<span><span class="co">## Trainable params: 998 (3.90 KB)</span></span> |
|
|
218 |
<span><span class="co">## Non-trainable params: 16 (64.00 Byte)</span></span> |
|
|
219 |
<span><span class="co">## _________________________________________________________________</span></span></code></pre> |
|
|
220 |
<p>Finally we can train the model.</p> |
|
|
221 |
<div class="sourceCode" id="cb11"><pre class="downlit sourceCode r"> |
|
|
222 |
<code class="sourceCode R"><span><span class="va">hist</span> <span class="op"><-</span> <span class="fu"><a href="../reference/train_model.html">train_model</a></span><span class="op">(</span><span class="va">model</span>,</span> |
|
|
223 |
<span> train_type <span class="op">=</span> <span class="st">"label_header"</span>,</span> |
|
|
224 |
<span> run_name <span class="op">=</span> <span class="st">"gc_model_1"</span>,</span> |
|
|
225 |
<span> path <span class="op">=</span> <span class="va">train_dir</span>,</span> |
|
|
226 |
<span> path_val <span class="op">=</span> <span class="va">validation_dir</span>,</span> |
|
|
227 |
<span> epochs <span class="op">=</span> <span class="fl">5</span>,</span> |
|
|
228 |
<span> steps_per_epoch <span class="op">=</span> <span class="fl">2</span>, <span class="co"># 20</span></span> |
|
|
229 |
<span> batch_size <span class="op">=</span> <span class="fl">64</span>,</span> |
|
|
230 |
<span> step <span class="op">=</span> <span class="fl">50</span>, </span> |
|
|
231 |
<span> path_tensorboard <span class="op">=</span> <span class="va">tb_dir</span>, <span class="co"># path to tensorboard logs</span></span> |
|
|
232 |
<span> tb_images <span class="op">=</span> <span class="cn">F</span>, <span class="co"># show confusion matrix in tensorboard</span></span> |
|
|
233 |
<span> vocabulary_label <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="st">"high_gc"</span>, <span class="st">"equal_dist"</span><span class="op">)</span><span class="op">)</span></span> |
|
|
234 |
<span> </span> |
|
|
235 |
<span><span class="fu"><a href="https://rdrr.io/r/graphics/plot.default.html" class="external-link">plot</a></span><span class="op">(</span><span class="va">hist</span><span class="op">)</span></span></code></pre></div> |
|
|
236 |
<p>Use the following command to open tensorboard in a browser:</p> |
|
|
237 |
<div class="sourceCode" id="cb12"><pre class="downlit sourceCode r"> |
|
|
238 |
<code class="sourceCode R"><span><span class="fu">keras</span><span class="fu">::</span><span class="fu"><a href="https://rdrr.io/pkg/tensorflow/man/tensorboard.html" class="external-link">tensorboard</a></span><span class="op">(</span><span class="va">tb_dir</span><span class="op">)</span></span></code></pre></div> |
|
|
239 |
<p>We can observe the scores for loss, accuracy, balanced accuracy, F1 |
|
|
240 |
and AUC under the “SCALARS” tab.</p> |
|
|
241 |
<p><img src="tb_images/loss.png" height="300"></p> |
|
|
242 |
<p><img src="tb_images/acc.png" height="300"></p> |
|
|
243 |
<p><img src="tb_images/bal_acc.png" height="300"></p> |
|
|
244 |
<p><img src="tb_images/f1.png"></p> |
|
|
245 |
<p><img src="tb_images/auc.png" height="300"></p> |
|
|
246 |
<p>We can also observe how the learning rate might have changed</p> |
|
|
247 |
<p><img src="tb_images/lr.png" height="300"></p> |
|
|
248 |
<p>In the “training files seen” window, we can observe how often we |
|
|
249 |
iterated over the training files.</p> |
|
|
250 |
<p><img src="tb_images/files_seen.png" height="300"></p> |
|
|
251 |
<p>In the “IMAGES” tab we can see a confusion matrix for the train and |
|
|
252 |
validation scores for every epoch</p> |
|
|
253 |
<p><img src="tb_images/cm_train.png" height="300"></p> |
|
|
254 |
<p><img src="tb_images/cm_val.png" height="300"></p> |
|
|
255 |
<p>In the “HPARAM” tab you can see hyper parameters of each run</p> |
|
|
256 |
<p><img src="tb_images/hparam.png" height="300"></p> |
|
|
257 |
</main> |
|
|
258 |
</div> |
|
|
259 |
|
|
|
260 |
|
|
|
261 |
|
|
|
262 |
<footer><div class="pkgdown-footer-left"> |
|
|
263 |
<p>Developed by Philipp Münch, René Mreches, Martin Binder, Hüseyin Anil Gündüz, Xiao-Yin To, Alice McHardy.</p> |
|
|
264 |
</div> |
|
|
265 |
|
|
|
266 |
<div class="pkgdown-footer-right"> |
|
|
267 |
<p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.0.9.</p> |
|
|
268 |
</div> |
|
|
269 |
|
|
|
270 |
</footer> |
|
|
271 |
</div> |
|
|
272 |
|
|
|
273 |
|
|
|
274 |
|
|
|
275 |
|
|
|
276 |
|
|
|
277 |
</body> |
|
|
278 |
</html> |