-
Notifications
You must be signed in to change notification settings - Fork 0
/
cs8850_08_bayes.html
618 lines (556 loc) · 30.1 KB
/
cs8850_08_bayes.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0, maximum-scale=1.0, user-scalable=no">
<link href="css/fontawesome-free-6.2.1-web/css/all.css" rel="stylesheet">
<!-- <link href="https://stackpath.bootstrapcdn.com/font-awesome/4.7.0/css/font-awesome.min.css" rel="stylesheet"/> -->
<script src="lib/colorbrewer.v1.min.js" charset="utf-8"></script>
<script src="lib/colorStringStandalone.js" charset="utf-8"></script>
<script type="text/javascript" src="lib/jquery-2.2.4.min.js"></script>
<title>Advanced Machine Learning</title>
<meta name="description" content="CS8850 GSU class">
<meta name="author" content="Sergey M Plis">
<meta name="apple-mobile-web-app-capable" content="yes">
<meta name="apple-mobile-web-app-status-bar-style" content="black-translucent">
<link rel="stylesheet" href="dist/reset.css">
<link rel="stylesheet" href="dist/reveal.css">
<!-- Code syntax highlighting -->
<link rel="stylesheet" href="plugin/highlight/monokai.css" id="highlight-theme">
<!-- <link rel="stylesheet" href="lib/css/zenburn.css"> -->
<link rel="stylesheet" href="css/custom.css">
<link rel="stylesheet" href="dist/theme/aml.css" id="theme">
<!-- Printing and PDF exports -->
<script>
var link = document.createElement( 'link' );
link.rel = 'stylesheet';
link.type = 'text/css';
link.href = window.location.search.match( /print-pdf/gi ) ? 'css/print/pdf.css' : 'css/print/paper.scss';
document.getElementsByTagName( 'head' )[0].appendChild( link );
</script>
</head>
<body>
<div class="reveal">
<!-- In between the <div="reveal"> and the <div class="slides">-->
<!-- <header style="position: absolute; top: 10px; left: 100px; z-index: 500; font-size:100px;background-color: rgba(0,0,0,0); text-align: center !important"></header> -->
<!-- In between the <div="reveal"> and the <div class="slides">-->
<!-- Any section element inside of this container is displayed as a slide -->
<div class="slides">
<section>
<section>
<p>
<h2>Advanced Machine Learning</h2>
<h3>08: Bayesian Decision Theory</h3>
<p>
</section>
<section>
<h3>Schedule</h3>
<row>
<col50>
<table style="font-size:14px">
<tr>
<th>#</th>
<th>date</th>
<th>topic</th>
<th>description</th>
</tr>
<tr><td>1</td>
<td> 22-Aug-2022 </td>
<td> Introduction </td>
<td></td>
</tr>
<tr>
<td> 2 </td>
<td> 24-Aug-2022 </td>
<td> Foundations of learning </td>
<td> </td>
</tr>
<tr><td> 3 </td><td> 29-Aug-2022 </td><td> PAC learnability </td><td> </td></tr>
<tr><td> 4 </td><td> 31-Aug-2022 </td><td> Linear algebra (recap) </td><td> hw1 released </td></tr>
<tr style='background-color: #FBEEC2;'><td> </td><td> 05-Sep-2022 </td><td> <em>Holiday</em> </td><td> </td></tr>
<tr style='background-color: #E0E4CC;'><td> 5 </td><td> 07-Sep-2022 </td><td> Linear learning models </td><td> </td></tr>
<tr><td> 6 </td><td> 12-Sep-2022 </td><td> Principal Component Analysis </td><td> project ideas </td></tr>
<tr><td> 7 </td><td> 14-Sep-2022 </td><td> Curse of Dimensionality </td></td></td><td> hw1 due </td></tr>
<tr><td> 8 </td><td> 19-Sep-2022 </td><td> Bayesian Decision Theory </td><td>hw2 release <i class='fa fa-map-marker' style='color: #FA6900;'></i> </td></tr>
<tr><td> 9 </td><td> 21-Sep-2022 </td><td> Parameter estimation: MLE </td><td></td></tr>
<tr><td> 10 </td><td> 26-Sep-2022 </td><td> Parameter estimation: MAP & NB</td><td>finalize teams</td></tr>
<tr><td> 11 </td><td> 28-Sep-2022 </td><td> Logistic Regression </td><td> </td></tr>
<tr><td> 12 </td><td> 03-Oct-2022 </td><td> Kernel Density Estimation </td><td> </td></tr>
<tr><td> 13 </td><td> 05-Oct-2022 </td><td> Support Vector Machines </td><td> hw3, hw2 due </td></tr>
<tr style='background-color: #E5DDCB;'><td> </td><td> 10-Oct-2022 </td><td> * Mid-point projects checkpoint </td><td> * </td></tr>
<tr style='background-color: #E5DDCB;'><td> </td><td> 12-Oct-2022 </td><td> * Midterm: Semester Midpoint </td><td> exam </td></tr>
<tr><td> 14 </td><td> 17-Oct-2022 </td><td>Matrix Factorization</td><td> </td></tr>
<tr><td> 15 </td><td> 19-Oct-2022 </td><td>Stochastic Gradient Descent</td><td> </td></tr>
</table>
</col50>
<col50>
<table style="font-size:14px; vertical-align: top;">
<tr>
<th>#</th>
<th>date</th>
<th>topic</th>
<th>description</th>
</tr>
<tr><td> 16 </td><td> 24-Oct-2022 </td><td> k-means clustering </td><td> </td></tr>
<tr><td> 17 </td><td> 26-Oct-2022 </td><td> Expectation Maximization </td><td> hw4, hw3 due </td></tr>
<tr><td> 18 </td><td> 31-Oct-2022 </td><td> Automatic Differentiation </td><td> </td></tr>
<tr><td> 19 </td><td> 02-Nov-2022 </td><td> Nonlinear embedding approaches </td><td> </td></tr>
<tr><td> 20 </td><td> 07-Nov-2022 </td><td> Model comparison I </td><td> </td></tr>
<tr><td> 21 </td><td> 09-Nov-2022 </td><td> Model comparison II </td><td> hw5, hw4 due</td></tr>
<tr><td> 22 </td><td> 14-Nov-2022 </td><td> Model Calibration </td><td> </td></tr>
<tr><td> 23 </td><td> 16-Nov-2022 </td><td> Convolutional Neural Networks </td><td> </td></tr>
<tr style='background-color: #FBEEC2;'><td> </td><td> 21-Nov-2022 </td><td> <em>Fall break</em> </td><td> </td></tr>
<tr style='background-color: #FBEEC2;'><td> </td><td> 23-Nov-2022 </td><td> <em>Fall break</em> </td><td> </td></tr>
<tr><td> 24 </td><td> 28-Nov-2022 </td><td> Word Embedding </td><td> hw5 due </td></tr>
<tr style='background-color: #FBEEC2;'><td> </td><td> 30-Nov-2022 </td><td> Presentation and exam prep day </td><td> </td></tr>
<tr style='background-color: #E5DDCB;'><td> </td><td> 02-Dec-2022 </td><td> * Project Final Presentations </td><td> * </td></tr>
<tr style='background-color: #E5DDCB;'><td> </td><td> 07-Dec-2022 </td><td> * Project Final Presentations </td><td> * </td></tr>
<tr style='background-color: #E5DDCB;'><td> </td><td> 12-Dec-2022 </td><td> * Final Exam </td><td> * </td></tr>
<tr><td> </td><td> 15-Dec-2022 </td><td> Grades due </td><td> </td></tr>
</table>
</col50>
</row>
</section>
<section>
<h3>Outline for the lecture</h3>
<ul>
<li class="fragment roll-in"> Preliminaries
<li class="fragment roll-in"> The Bayesian View
<li class="fragment roll-in">
</ul>
</section>
</section>
<!-- ------------------------------------------------------------------------- -->
<section>
<section>
<h2>Preliminaries</h2>
<blockquote style="background-color: #93a1a1; color: #fdf6e3; font-size: 38px; width=100%; text-align: justify;">
Probability theory is nothing but common sense
reduced to calculation<br>
- Pierre-Simon Laplace, 1812
</blockquote>
</section>
<section>
<h3>Deductive and plausible reasoning</h3>
<aside class="notes">
talk about
</aside>
</section>
<section data-background="figures/Argument_terminology_used_in_logic.svg" data-background-size="contain">
<aside class="notes">
talk about
</aside>
</section>
<section data-background="figures/BN_animated.gif" data-background-size="contain">
<aside class="notes">
talk about
</aside>
</section>
<section data-background="figures/risk-simple.gif" data-background-size="contain">
<aside class="notes">
talk about
</aside>
</section>
<section>
<h3>Random Variables</h3>
<aside class="notes">
</aside>
</section>
<section>
<h3>Probability Distributions</h3>
<aside class="notes">
</aside>
</section>
<section>
<h3>Probability Distributions</h3>
<aside class="notes">
</aside>
</section>
</section>
<!-- ------------------------------------------------------------------------- -->
<section>
<section>
<h2>The Bayesian view</h2>
<div class="slide-footer">
closely follows Duda, R. O., Hart, P. E., & Stork, D. G. (2012). Pattern classification. John Wiley & Sons.
</div>
</section>
<section>
<h2>Bayesian decision theory</h2>
<blockquote>
<b>Definition:</b> Quantify the trade-offs
between various classification decisions based
on probabilities and the costs that accompany
such decisions
</blockquote>
<b>Assumptions:</b>
<ul>
<li class="fragment roll-in"> The decision problem is posed in probabilistic terms
<li class="fragment roll-in"> All relevant probability values are known
</ul>
</section>
<section data-background="figures/fish_sorting.gif" data-state="header1">
<style>.header1 header:after { content: "Fish sorting machine"; }</style>
</section>
<section>
<h2>The Prior</h2>
<ul>
<li class="fragment roll-in"> $\omega$ is a random variable: fish type
<li class="fragment roll-in"> $\omega\in \{\omega_1, \omega_2\}$ $\omega_1$ - salmon, $\omega_2$ - sea bass
<li class="fragment roll-in"> $\prob{P}{\omega_1}$ - the <em>a priory</em> probability (prior) that fish is a salmon
<li class="fragment roll-in"> $\prob{P}{\omega_2}$ - the <em>a priory</em> probability (prior) that fish is a sea bass
<li class="fragment roll-in">
<blockquote>
Gives us the knowledge of how likely we are to get salmon before we see any fish
</blockquote>
</ul>
</section>
<section>
<h2>Priors-only decision rule</h2>
<blockquote>
How do we make a decision between $\omega_1$ and $\omega_2$ if all we know is the priors $\prob{P}{\omega_1}$ and $\prob{P}{\omega_2}$?
</blockquote>
<div class="fragment" data-fragment-index="0" >
<ul style="list-style-type: none;">
<li class="fragment roll-in">
<table>
<tr>
<th>decide</th>
<th>if</th>
</tr>
<tr>
<td>salmon</td>
<td>$\prob{P}{\omega_1} \gt \prob{P}{\omega_2}$</td>
</tr>
<tr>
<td>sea bass</td>
<td>$\prob{P}{\omega_1} \lt \prob{P}{\omega_2}$</td>
</tr>
</table>
<li class="fragment roll-in"> Thus minimize your probability of being incorrect
</ul>
</div>
</section>
<section>
<h3>$\prob{P}{\mbox{meet}} \gt \prob{P}{\overline{\mbox{meet}}}\rightarrow\mbox{shoot}$</h3>
<img style="border:0; box-shadow: 0px 0px 0px rgba(255, 255, 255, 255);" width="900" class="reveal"
src="figures/bart.jpg" alt="holistic assassin">
</section>
<section>
<h2>Decide based just on the prior</h2>
<ul style="list-style-type: none;">
<li class="fragment roll-in"> If $\prob{P}{\omega_1} \gg \prob{P}{\omega_2}$, you are right most of the time when deciding $\omega_1$
<li class="fragment roll-in"> If $\prob{P}{\omega_1} = \prob{P}{\omega_2}$, you are taking a random guess when deciding $\omega_1$
<li class="fragment roll-in"> No other decision rule can yield larger (expected) probability of being right
</ul>
</section>
<section>
<h2>Use observations to increase our chances</h2>
<div class="row">
<div class="col_left5">
<ul>
<li> If we know the <em>class conditional probability density</em> $\prob{P}{\vec{x}|\omega_i}$ we can make better decisions
<li> For example, $\vec{x}$ is the observed lightness of a fish
</ul>
</div>
<div class="col_right">
<img style="border:0; box-shadow: 0px 0px 0px rgba(150, 150, 255, 1);" width="1000"
src="figures/likelihood.png" alt="likelihood">
</div>
</div>
</section>
<section>
<div id="header-right" style="right: -10%;">
<img width="200px" style="margin-bottom: -5%"
src="figures/bayes.png" alt="rev. Bayes"><br>
<small>Thomas Bayes</small>
</div>
<h2>Bayes theorem</h2>
<ul>
<li class="fragment roll-in"> Use the observed feature(s) $\vec{x}$ and the Bayes theorem
<blockquote style="background-color: #eee8d5; font-size: 28px;" >
\[
\prob{P}{\omega_i|\vec{x}} = \frac{\prob{P}{\vec{x}|\omega_i}\prob{P}{\omega_i}}{\prob{P}{\vec{x}}}
\]
</blockquote>
<li class="fragment roll-in"> In "plain English"
<blockquote style="background-color: #eee8d5; font-size: 28px;">
\[
\mbox{posterior} = \frac{\mbox{likelihood}\times\mbox{prior}}{\mbox{evidence}}
\]
</blockquote>
<li class="fragment roll-in"> Evidence for $C$ classes
$
\prob{P}{\vec{x}} = \sum_i^C\prob{P}{\vec{x}|\omega_i}\prob{P}{\omega_i}
$
</ul>
<aside class="notes">
explain how we got there
</aside>
</section>
<section data-fullscreen>
<div id="header-right" style="right: -0%;">
<div style="font-size:22px">
\[
\prob{P}{\omega_i|\vec{x}} = \frac{\prob{P}{\vec{x}|\omega_i}\prob{P}{\omega_i}}{\prob{P}{\vec{x}}}
\]
</div>
</div>
<h2>Put this to numbers</h2>
<blockquote style="text-align: left; background-color: #eee8d5; font-size:30px;">
One person in 200,000 has a particular
Progressive multifocal leukoencephalopathy
(PML). There is a diagnostic test for the
disease. It is correct 99% of time. Your test is
positive what's the probability you have PML?
</blockquote>
<row>
<col50 style="font-size:20pt;">
<ul>
<li class="fragment roll-in"> $\vec{x}$ - tested positively
<li class="fragment roll-in"> $\prob{P}{+|PML} = 0.99$
<li class="fragment roll-in"> $\prob{P}{+|\overline{PML}} = 0.01$
<li class="fragment roll-in"> $\prob{P}{PML} = \frac{1}{200,000}$
<li class="fragment roll-in"> $\prob{P}{\overline{PML}} = \frac{199999}{200,000}$
<li class="fragment roll-in"> $\prob{P}{PML|+} \approx 0.05\%$
</ul>
</col50>
<col50 style="font-size:20pt; text-align: left;">
\begin{align}
\prob{P}{+} & = \prob{P}{+|PML}\prob{P}{PML}\\
& + \prob{P}{+|\overline{PML}}\prob{P}{\overline{PML}}\\
\prob{P}{PML|+} &= \frac{\prob{P}{+|PML}\prob{P}{PML}}{\prob{P}{+}}
\end{align}
</col50>
</row>
</section>
<section>
<h3>Decide based on posterior</h3>
<div class="row">
<div class="col_right">
<ul style="list-style-type: none; font-size: 22pt">
<li class="fragment roll-in"> If $\prob{P}{\omega_1|\vec{x}} \gt \prob{P}{\omega_2|\vec{x}}$, decide $\omega_1$
<li class="fragment roll-in"> If $\prob{P}{\omega_1|\vec{x}} \lt \prob{P}{\omega_2|\vec{x}}$, decide $\omega_2$
<li class="fragment roll-in"> $\prob{P}{error|\vec{x}} = \min[\prob{P}{\omega_1|\vec{x}}, \prob{P}{\omega_2|\vec{x}}]$
</ul>
</div>
<div class="col_left5">
<img style="border:0; box-shadow: 0px 0px 0px rgba(150, 150, 255, 1);" width="1000"
src="figures/posterior_ratio.svg" alt="posterior">
</div>
</section>
<section>
<h3>General Bayesian Decision Theory</h3>
<ul>
<li class="fragment roll-in"> Multiple features in $\vec{x}$
<li class="fragment roll-in"> More than two states of nature (classes)
<li class="fragment roll-in"> Allow actions (also refusal to act)
<li class="fragment roll-in"> Loss function more general that probability of error
</ul>
</section>
<section>
<h3>General Bayesian Decision Theory: formally</h3>
<ul style="line-height:100%;">
<li class="fragment roll-in"> $\Omega = \{ \omega_1, \omega_2, \dots, \omega_C\}$ $c$ categories
<li class="fragment roll-in"> $\vec{x} = \{x_1, x_2, \dots, x_d\}$ $d$-dimensional feature vector
<li class="fragment roll-in"> $\vec{\alpha} = \{\alpha_1, \alpha_2, \dots, \alpha_A\}$ set of actions
<li class="fragment roll-in"> $\lambda(\alpha_i|\omega_i)$ loss incurred for taking action $\alpha_i$ in case of $\omega_i$
<li class="fragment roll-in"> Conditional risk
\begin{array}
\prob{R}{\alpha_i|\vec{x}} = \sum_j^C \lambda(\alpha_i|\omega_j)\prob{P}{\omega_j|\vec{x}}
\end{array}
<li class="fragment roll-in" style="font-size: 22pt;"> Total risk
\begin{array}
\prob{R}{\vec{x}} = \sum_i^A \prob{R}{\alpha_i|\vec{x}}
\end{array}
</ul>
</section>
<section>
<div id="header-right">
<div style="font-size:22px">
\[
\prob{N}{\mu, \sigma} = \frac{1}{\sigma\sqrt{2\pi}} e^{ -\frac{1}{2}\left(\frac{x-\mu}{\sigma}\right)^2 }
\]
</div>
</div>
<h3>Assignment</h3>
<blockquote>
Draw probability densities and find the decision regions for the following classes:
</blockquote>
<ul style="list-style-type: none; font-size: 22pt; ">
<li> $\Omega = \{ \omega_1, \omega_2\}$
<li> $\prob{P}{x|\omega_1} \sim \prob{N}{20, 4}$
<li> $\prob{P}{x|\omega_2} \sim \prob{N}{15, 2}$
<li> $\prob{P}{\omega_1} = \frac{1}{3}$
<li> $\prob{P}{\omega_2} = \frac{2}{3}$
<li> Classify $x = 17$
</ul>
</section>
<section>
<h2>Maximum A Posteriori</h2>
<h3>The poor man's Bayes</h3>
<div class="fragment" data-fragment-index="0">
<img style="border:0; box-shadow: 0px 0px 0px rgba(255, 255, 255, 255);" width="1000"
src="figures/beta_concentrate.svg" alt="beta">
$n = \alpha_H + \alpha_T$ increases $\rightarrow$
</div>
<ul>
<li class="fragment roll-in"> Potentially, can know all there's to know about posterior
<li class="fragment roll-in"> But chose to go for a single parameter estimate
</ul>
<aside class="notes">
is just estimating a point parameter - the maximum of posterior<br>
kinda not fully Bayesian<br>
The prior and posterior are in the parameter domain<br>
</aside>
</section>
<section>
<h2>Can estimate full distributions</h2>
<h3>need to resort to sampling</h3>
<div class="fragment" data-fragment-index="0">
<img style="border:0; box-shadow: 0px 0px 0px rgba(255, 255, 255, 255);" width="700"
src="figures/posterior_smapling.png" alt="sampling">
</div>
<aside class="notes">
Bayesian approach can estimate distributions<br>
If we construct our posterior using a chain rule, we can sample from the components and generate samples from the posterior <br>
</aside>
</section>
<section>
<h2>Choosing conjugate priors</h2>
<h3>striving for simple analytical forms</h3>
<div class="fragment" data-fragment-index="0">
<img style="border:0; box-shadow: 0px 0px 0px rgba(255, 255, 255, 255);" width="600"
src="figures/Dirichlet-3d-panel.png" alt="Dirichlet">
</div>
<aside class="notes">
Prior defines a distribution over likelihoods - also distributions<br>
Sample hyperparameters - get a newdistribution
</aside>
</section>
</section>
</div>
</div>
<script src="dist/reveal.js"></script>
<link rel="stylesheet" href="plugin/highlight/monokai.css">
<script src="plugin/highlight/highlight.js"></script>
<script src="plugin/math/math.js"></script>
<script src="plugin/chalkboard/plugin.js"></script>
<script src="plugin/notes/notes.js"></script>
<script src="plugin/zoom/zoom.js"></script>
<script src="plugin/fullscreen/fullscreen.js"></script>
<script src="plugin/menu/menu.js"></script>
<script>
// Full list of configuration options available at:
// https://github.com/hakimel/reveal.js#configuration
Reveal.initialize({
// history: true,
hash: true,
margin: 0.01,
minScale: 0.01,
maxScale: 1.23,
menu: {
themes: false,
openSlideNumber: true,
openButton: false,
},
chalkboard: {
boardmarkerWidth: 1,
chalkWidth: 2,
chalkEffect: 1,
toggleNotesButton: false,
toggleChalkboardButton: false,
slideWidth: Reveal.width,
slideHeight: Reveal.height,
// src: "chalkboards/chalkboard_em2.json",
readOnly: false,
theme: "blackboard",
eraser: { src: "plugin/chalkboard/img/sponge.png", radius: 30},
},
math: {
mathjax: 'https://cdn.jsdelivr.net/gh/mathjax/[email protected]/MathJax.js',
config: 'TeX-AMS_SVG-full',
// pass other options into `MathJax.Hub.Config()`
TeX: {
Macros: {
RR: '\\mathbb{R}',
PP: '\\mathbb{P}',
EE: '\\mathbb{E}',
NN: '\\mathbb{N}',
vth: '\\vec{\\theta}',
loss: '{\\cal l}',
hclass: '{\\cal H}',
CD: '{\\cal D}',
def: '\\stackrel{\\text{def}}{=}',
pag: ['\\text{pa}_{{\cal G}^{#1}}(#2)}', 2],
vec: ['\\boldsymbol{\\mathbf #1}', 1],
set: [ '\\left\\{#1 \\; : \\; #2\\right\\}', 2 ],
bm: ['\\boldsymbol{\\mathbf #1}', 1],
argmin: ['\\operatorname\{arg\\,min\\,\}'],
argmax: ['\\operatorname\{arg\\,max\\,\}'],
prob: ["\\mbox{#1$\\left(#2\\right)$}", 2],
},
loader: {load: ['[tex]/color']},
extensions: ["color.js"],
tex: {packages: {'[+]': ['color']}},
svg: {
fontCache: 'global'
}
}
},
plugins: [ RevealMath, RevealChalkboard, RevealHighlight, RevealNotes, RevealZoom, RevealMenu ],
});
Reveal.configure({ fragments: true }); // set false when developing to see everything at once
Reveal.configure({ slideNumber: true });
//Reveal.configure({ history: true });
Reveal.configure({ slideNumber: 'c / t' });
Reveal.addEventListener( 'darkside', function() {
document.getElementById('theme').setAttribute('href','dist/theme/aml_dark.css');
}, false );
Reveal.addEventListener( 'brightside', function() {
document.getElementById('theme').setAttribute('href','dist/theme/aml.css');
}, false );
</script>
<style type="text/css">
/* 1. Style header/footer <div> so they are positioned as desired. */
#header-left {
position: absolute;
top: 0%;
left: 0%;
}
#header-right {
position: absolute;
top: 0%;
right: 0%;
}
#footer-left {
position: absolute;
bottom: 0%;
left: 0%;
}
</style>
<!-- // 2. Create hidden header/footer -->
<div id="hidden" style="display:none;">
<div id="header">
<div id="header-left"><h4>CS8850</h4></div>
<div id="header-right"><h4>Advanced Machine Learning</h4></div>
<div id="footer-left">
<img style="border:0; box-shadow: 0px 0px 0px rgba(150, 150, 255, 1);" width="200"
src="figures/valentino.png" alt="robot learning">
</div>
</div>
</div>
<script type="text/javascript">
// 3. On Reveal.js ready event, copy header/footer <div> into each `.slide-background` <div>
var header = $('#header').html();
if ( window.location.search.match( /print-pdf/gi ) ) {
Reveal.addEventListener( 'ready', function( event ) {
$('.slide-background').append(header);
});
}
else {
$('div.reveal').append(header);
}
</script>
</body>
</html>