-
Notifications
You must be signed in to change notification settings - Fork 1
/
index.xml
199 lines (199 loc) · 36.2 KB
/
index.xml
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
<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Amirmohammad Kazemeini's Blog</title><link>https://amirmohammadkz.github.io/</link><atom:link href="https://amirmohammadkz.github.io/index.xml" rel="self" type="application/rss+xml"/><description>Amirmohammad Kazemeini's Blog</description><generator>Wowchemy (https://wowchemy.com)</generator><language>en-us</language><lastBuildDate>Thu, 01 Sep 2022 00:00:00 +0000</lastBuildDate><image><url>https://amirmohammadkz.github.io/images/icon_hu0b7a4cb9992c9ac0e91bd28ffd38dd00_9727_512x512_fill_lanczos_center_3.png</url><title>Amirmohammad Kazemeini's Blog</title><link>https://amirmohammadkz.github.io/</link></image><item><title>Psychological Understanding of Textual journals using Natural Language Processing approaches</title><link>https://amirmohammadkz.github.io/publication/thesis/</link><pubDate>Thu, 01 Sep 2022 00:00:00 +0000</pubDate><guid>https://amirmohammadkz.github.io/publication/thesis/</guid><description/></item><item><title>Interpretable Representation Learning for Personality Detection</title><link>https://amirmohammadkz.github.io/publication/icdm2021/</link><pubDate>Sun, 10 Jan 2021 00:00:00 +0000</pubDate><guid>https://amirmohammadkz.github.io/publication/icdm2021/</guid><description/></item><item><title>Multitask Learning for Emotion and Personality Detection</title><link>https://amirmohammadkz.github.io/publication/mtl/</link><pubDate>Sun, 10 Jan 2021 00:00:00 +0000</pubDate><guid>https://amirmohammadkz.github.io/publication/mtl/</guid><description/></item><item><title>If I was a machine, part 1: The future may not be as beautiful as we think.</title><link>https://amirmohammadkz.github.io/post/post1/</link><pubDate>Sun, 27 Sep 2020 00:00:00 +0000</pubDate><guid>https://amirmohammadkz.github.io/post/post1/</guid><description><p>Hello, this is the first part of this serie. In this story, I try to question some of the dark sides of AI, which I think are forgotten or neglected.</p>
<h3 id="why-should-you-read-this">Why should you read this?</h3>
<p>This story is not a state-of-the-art paper explanation, nor is it a pure academic article. As a recent graduate research student learning NLP to leverage it in psychology and social science, I will share my thoughts with you. I believe that we are inherently creative, and the structure of learning, and the way we restrict ourselves in the systems, will deteriorate it. Inevitably, I should follow a similar path as other students did, and I think my creative aspect will be diminished because the whole process is a spoiler path. We are bombarded with pure knowledge without finding out what the problem is and how we can invent a solution. Limited in the fixed study routines, trying to read all the requirements of a problem, and then we are expected to design and create after it. But how can we do it? This process, each step of this trail, contains a host of opportunities for creating. However, we are obligated to study all the requirements first because the chance of inventing something that is not made before and performs better is not high. It might be right, but we are improving this chance in the cost of sacrificing creativity, the essential tool for producing novelty. So, I decided not to follow the typical path and make my way a bit slower. I try to do this by guessing the next step each time I learn a new subject, questioning the most prominent drawbacks of current methods before jumping into state-of-the-art discussions from the most famous researchers.
By following these steps, I may go outside of the box, because I do not know where the box is now, that is the key. When I am aware of the box, outside can be interpreted as a new box, “the box of outside”. Maybe the first chapters are a bit rusty and unstructured, but I try to improve it bit by bit. So, if you agree with me, follow my journey.</p>
<h2 id="nlp-and-ai-are-against-morality">NLP and AI are against morality</h2>
<figure id="figure-two-main-characters-kogami-and-mikishima-in-psycho-pass-photo-from-wallpaperaccesshttpswallpaperaccesscompsycho-pass">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/psycho-pass.jpg" data-caption="Two Main Characters (Kogami and Mikishima) in Psycho Pass. Photo from &lt;a href=&#34;https://wallpaperaccess.com/psycho-pass&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;wallpaperaccess&lt;/a&gt;">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/psycho-pass.jpg" alt="" >
</a>
<figcaption>
Two Main Characters (Kogami and Mikishima) in Psycho Pass. Photo from <a href="https://wallpaperaccess.com/psycho-pass" target="_blank" rel="noopener">wallpaperaccess</a>
</figcaption>
</figure>
<p>Yeah, it may seem weird to hear that from an NLP student, but unfortunately, at least in some parts, it is true. I know it depends on how I define morality, but I think you may agree with me in the general case. It first came into my mind when I was watching <a href="https://en.wikipedia.org/wiki/Psycho-Pass" target="_blank" rel="noopener">‘psycho pass’</a> anime. In the first minutes of it, or by just reading the summary before watching, you will understand that it is about a society that is precisely monitored by AI. Identifying potential criminals, and preventing the possible crimes, predicting the proper job for everyone, etc. I do not want to spoil the serie, and I do not know how it will end because the rest of it is under production. But by just knowing some of these concepts, some questions triggered in my mind. I asked <a href="https://dirkhovy.com/" target="_blank" rel="noopener">Dirk Hovy</a> , the famous researcher in the field of ethics in AI, about some of them at ACL 2020, and he agreed, and then I understood that these are concerning questions that can seriously affect our future. Let me tell you some of them:</p>
<h3 id="1-with-perfect-ai-models-hope-will-not-exist-anymore">1. With perfect AI models, hope will not exist anymore</h3>
<figure id="figure-photo-from-shutterstockhttpswwwshutterstockcom">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/b3ae473bb7177446b2416b4e73f7d56ffb792763/static/media/hope.gif" data-caption="Photo from &lt;a href=&#34;https://www.shutterstock.com/&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;shutterstock&lt;/a&gt;">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/b3ae473bb7177446b2416b4e73f7d56ffb792763/static/media/hope.gif" alt="" >
</a>
<figcaption>
Photo from <a href="https://www.shutterstock.com/" target="_blank" rel="noopener">shutterstock</a>
</figcaption>
</figure>
<p>Unfortunately, that is true. Suppose a kid who is trying to be a literature teacher, but his talent is in mathematics. A perfect model will decide that he is expected to be a mathematician rather than a poet or a literature teacher. From a broader perspective, each individual might only have a few options to pursue to build the most productive community. Sadly, our models are not capable of understanding our feelings and senses. They merely can interpret them by using the way we express ourselves in a limited manner. This is not sufficient, nor is it proper for understanding someone’s inner feelings.</p>
<h3 id="2-no-one-will-think-outside-of-the-box-anymore">2. No one will think outside of the box anymore</h3>
<figure id="figure-photo-by-markus-spiskehttpsunsplashcommarkusspiskeutm_sourcemediumutm_mediumreferral-on-unsplashhttpsunsplashcom">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/4774499face7ac420ac86348714d6c47cfc94adb/static/media/box.jpg" data-caption="Photo by &lt;a href=&#34;https://unsplash.com/@markusspiske?utm_source=medium&amp;amp;utm_medium=referral&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;Markus Spiske&lt;/a&gt; on &lt;a href=&#34;https://unsplash.com/&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;Unsplash&lt;/a&gt;">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/4774499face7ac420ac86348714d6c47cfc94adb/static/media/box.jpg" alt="" >
</a>
<figcaption>
Photo by <a href="https://unsplash.com/@markusspiske?utm_source=medium&amp;utm_medium=referral" target="_blank" rel="noopener">Markus Spiske</a> on <a href="https://unsplash.com/" target="_blank" rel="noopener">Unsplash</a>
</figcaption>
</figure>
<p>Creativity is a common phenomenon that we realize is much more than just producing some apparently new things by combining the available ones. <a href="https://en.wikipedia.org/wiki/Generative_adversarial_network" target="_blank" rel="noopener">GAN</a> and <a href="https://en.wikipedia.org/wiki/Deepfake" target="_blank" rel="noopener">DeepFake</a> models are trying to do so, as well as <a href="https://en.wikipedia.org/wiki/GPT-3" target="_blank" rel="noopener">GPT-3</a> for Natural Language Generation. Suppose a sage who knows everything on the internet. Does a generative model like GPT-3 produce a novel article or even a paragraph that surprises the sage? The answer is no. Because everything it generates is somehow available on the internet. You first may become amazed by reading the Guardian’s <a href="https://www.theguardian.com/commentisfree/2020/sep/08/robot-wrote-this-article-gpt-3" target="_blank" rel="noopener">“why humans have nothing to fear from AI”</a> article . But the truth is that the model just decides which concepts and words are appropriate to be written by remembering the memorized texts. If you disagree, have a look at this <a href="https://www.youtube.com/watch?v=SY5PvZrJhLE" target="_blank" rel="noopener">scientific explanation</a> by Yannic Kilcher. Some new methods such as <a href="https://sentic.net/senticnet-6.pdf" target="_blank" rel="noopener">SenticNet</a>, tried to combine human logic into subsymbolic AI, to make models that understand more and can predict better in the way we think. The idea is inspired by this wise saying:</p>
<p><strong>“you do not get discoveries in the sciences by taking huge amounts of data, throwing them into a computer and doing statistical analysis of them: that’s not the way you understand things, you have to have theoretical insights.”</strong> <em>Noam Chomsky</em></p>
<p>Despite all these efforts, the problem has not been solved yet. You may think that this limitation is not a significant concern. But as we go further, the situation becomes more serious. Our models will become more accurate, and then they will monitor our activities. Imagine the case that you write a pure novel book, completely different from all the previous ones. Models will now evaluate your manuscript to identify where should it deserve to be published. Since your work is unalike from the prior successful publications, it probably will not get accepted as valuable work. Hence, your creativity will be suppressed. All of this is because it does not understand more than what has been learned before. It cannot digest the creativity.</p>
<p><strong>“We shape our tools and, thereafter, our tools shape us”</strong> <em>Marshall McLuhan</em></p>
<h3 id="3-freedom-in-a-cage-the-myth-of-evolution">3. Freedom in a cage, the myth of evolution</h3>
<figure id="figure-freedom-cage-published-november-27-2012-by-sherif-arafa-politicalcartoonscom">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/cage.png" data-caption="Freedom Cage published November 27, 2012 by Sherif Arafa politicalcartoons.com">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/cage.png" alt="" >
</a>
<figcaption>
Freedom Cage published November 27, 2012 by Sherif Arafa politicalcartoons.com
</figcaption>
</figure>
<p>This is one of the consequences of the previous topic. Let me rephrase the quote above: They shape our tools; thereafter, our tools govern us. That is the fact. Keep in mind that your current wellbeing is for the sake of previous revolutions, people who tried to change the situation to make a better world. When censorship became automatic, by using methods such as Deep Packet Inception, nobody will be free anymore. This is not only the creativity that is suppressed but also the freedom, and everything that endangers the benefits of current government will be sacrificed. Rulers become cruel, and they eliminate everything that threatens their dictatorship. That is not a complex problem because the models magnify their acts, and they only should be concern about reducing False-negative predictions, individuals who are capable of rebellion and labeled as innocent. ‘Recall’ should be high, and anything else does not matter. If we reach this state, the evolution of society will be halted.</p>
<h3 id="4-the-distance-between-social-classes-will-be-fixed-or-gradually-widen">4. The distance between social classes will be fixed, or gradually widen.</h3>
<figure id="figure-districts-of-panem-photo-from-businessinsiderhttpswwwbusinessinsidercom10-things-you-need-to-know-about-the-hunger-games-2012-3internationaltruerusirtthe-characters-live-in-districts-4">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/penam.gif" data-caption="Districts of Panem. Photo from &lt;a href=&#34;https://www.businessinsider.com/10-things-you-need-to-know-about-the-hunger-games-2012-3?international=true&amp;amp;r=US&amp;amp;IR=T#the-characters-live-in-districts-4&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;businessinsider&lt;/a&gt;">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/penam.gif" alt="" >
</a>
<figcaption>
Districts of Panem. Photo from <a href="https://www.businessinsider.com/10-things-you-need-to-know-about-the-hunger-games-2012-3?international=true&amp;r=US&amp;IR=T#the-characters-live-in-districts-4" target="_blank" rel="noopener">businessinsider</a>
</figcaption>
</figure>
<p>This heading is partially connected with the first one. The problem arises when you are restricted to perform specific tasks based on your background. If we scrutinize this issue, the tasks are chosen based on our ancestors’ previous successful experience. It means, if you were born in a wealthy family, higher-quality jobs are probable to be assigned to you, and if your family is not wealthy, you might not have a chance to improve it. Rich people become richer and poor ones become poorer. This issue can also be extended to other areas such as social classes because all these problems are somehow intertwined. This social class division is demonstrated in <a href="https://en.wikipedia.org/wiki/The_Hunger_Games" target="_blank" rel="noopener">The Hunger Games</a> series of novels and movies. This issue encompasses not only individuals but also communities. Each society section will become isolated, and all unnecessary interactions between districts might be eliminated. And a day may come that you will not understand situations and things beyond your own community, and you may not know that they exist.</p>
<h2 id="conclusion">Conclusion</h2>
<p>This post may be somehow pessimistic, somehow far from reality, but anyway, these are concerns that I think this issue deserves more attention. Some of them may be wrong, but I believe devoting some time to think about our field’s probable destination, even if it might be wrong, is necessary.</p>
<p>If you disagree with me or have recommendations and comments, I appreciate if you reach out to me.</p>
<h3 id="did-you-find-this-page-helpful-consider-sharing-it-">Did you find this page helpful? Consider sharing it 🙌</h3></description></item><item><title>Bottom-up and top-down: Predicting personality with psycholinguistic and language model features</title><link>https://amirmohammadkz.github.io/publication/icdm2020/</link><pubDate>Thu, 24 Sep 2020 00:00:00 +0000</pubDate><guid>https://amirmohammadkz.github.io/publication/icdm2020/</guid><description/></item><item><title>IELTS Reading Band 9 Tips — An End to End Self-Study Guide</title><link>https://amirmohammadkz.github.io/post/post2/</link><pubDate>Sat, 12 Sep 2020 00:00:00 +0000</pubDate><guid>https://amirmohammadkz.github.io/post/post2/</guid><description><h3 id="introduction">Introduction</h3>
<p>IELTS, which stands for International English Language Testing System, is a standard international English test that assesses English skills in four sections reading, writing, listening, and speaking. Most international institutes and universities put an IELTS minimum requirement for their international applicants, so they become sure that no communication issues happen for non-English speakers after the recruitment process.</p>
<p>Last year was so tough for me. Struggling with the undergraduate thesis, working on a research paper, communicating with professors and universities for my graduate program, and taking required tests were intensely overwhelming. I took IELTS because it was compulsory for my graduate study, and I succeeded in achieving the overall band score of 7.5. The impressive part of my result was the reading section, which I hopefully got a full mark score in that. IELTS Reading consists of 3 sections, 40 questions, 60 minutes</p>
<ul>
<li>
<p>Each section contains one long text.</p>
</li>
<li>
<p>Texts are authentic and are taken from books, journals, magazines and newspapers. They have been written for a non-specialist audience and are on academic topics of general interest.</p>
</li>
<li>
<p>Texts are appropriate to, and accessible to, candidates entering undergraduate or postgraduate courses or seeking professional registration.</p>
</li>
<li>
<p>Texts range from the descriptive and factual to the discursive and analytical. Texts may contain non-verbal materials such as diagrams, graphs or illustrations. If texts contain technical terms, then a simple glossary is provided.</p>
</li>
<li>
<p>There are three texts and 40 questions. <a href="https://www.examenglish.com/IELTS/IELTS_Academic_reading.html" target="_blank" rel="noopener">[reference]</a></p>
</li>
</ul>
<p>I will explain some tips that I used in my self-study to achieve this score in the current post.</p>
<p>The critical factor that makes this post unique is the fast achieving self-study approach that I will explain. For my undergraduate thesis, which resulted in an <a href="http://www.winlp.org/winlp-2020-workshop" target="_blank" rel="noopener">ACL2020 WiNLP</a> paper, I used ML approaches to gain the best result to detect individuals’ personality traits based on their essays. You can read the final version of my paper <a href="http://www.winlp.org/wp-content/uploads/2020/final_papers/40_Paper.pdf" target="_blank" rel="noopener">here</a>. Since I was overwhelmed with several tasks , I did not find much time to focus on my IELTS study. So, I decided to leverage my ML skills to build a focused study plan to improve my IELTS result. Hopefully, it worked, and here, you can read my self-designed method.</p>
<p>This method is inspired by the Deep Learning models training process and consists of four parts: weight initialization, forward pass, backpropagation and error analysis, and optimization.</p>
<h3 id="the-start-line--weight-initialization">The Start Line — Weight Initialization</h3>
<figure id="figure-photo-by-bradenhttpsunsplashcombradencollumutm_sourcemediumutm_mediumreferral-collum-on-unsplashhttpsunsplashcomutm_sourcemediumutm_mediumreferral">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/run.jpeg" data-caption="Photo by &lt;a href=&#34;https://unsplash.com/@bradencollum?utm_source=medium&amp;amp;utm_medium=referral&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;Braden&lt;/a&gt; Collum on &lt;a href=&#34;https://unsplash.com/?utm_source=medium&amp;amp;utm_medium=referral&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;Unsplash&lt;/a&gt;">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/run.jpeg" alt="" >
</a>
<figcaption>
Photo by <a href="https://unsplash.com/@bradencollum?utm_source=medium&amp;utm_medium=referral" target="_blank" rel="noopener">Braden</a> Collum on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral" target="_blank" rel="noopener">Unsplash</a>
</figcaption>
</figure>
<p>When you want to learn a new subject, you should have a brief understanding of what you are going to do and how to start. That is the exact process that is used for training deep models too.</p>
<figure id="figure-photo-from-cambridgeorg">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/ielts15.jpg" data-caption="Photo from cambridge.org">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/ielts15.jpg" alt="" >
</a>
<figcaption>
Photo from cambridge.org
</figcaption>
</figure>
<p>First, pick the dataset you want to learn. There are numerous test packs for IELTS, but none are as genuine as the original <a href="https://ieltspracticeonline.com/download-all-cambridge-ielts-books-pdfaudio-1-14/" target="_blank" rel="noopener">Cambridge IELTS</a> test books. As far as I know, 15 versions have been published, and each year, a newer version comes out based on the recent actual tests.</p>
<p>The second essential criterion before the beginning is how to start it? How can you commence something that you do not have any idea about it? That is easy! Just use what you have known before from your prior successful test experience and try to reuse it. For example, I remembered that I used to read the questions first and highlight the keywords before jumping into the text for my undergraduate university entrance exam. If you think that you are totally lost, you can get some help from these free resources, but keep in mind not to spend much on them at this phase:</p>
<ul>
<li>
<p><a href="https://www.cambridge.org/us/cambridgeenglish/catalog/cambridge-english-exams-ielts/new-insight-ielts" target="_blank" rel="noopener">New insight to IELTS</a></p>
</li>
<li>
<p><a href="https://ieltsliz.com/ielts-reading-lessons-information-and-tips/" target="_blank" rel="noopener">IELTS Liz</a></p>
</li>
<li>
<p><a href="https://ielts-simon.com/ielts-help-and-english-pr/ielts-reading/" target="_blank" rel="noopener">IELTS Simon</a></p>
</li>
<li>
<p><a href="https://www.youtube.com/playlist?list=PLdawRnR9ilZDh4mWAjOm8MsqaJt03-YMR" target="_blank" rel="noopener">E2 IELTS</a></p>
</li>
</ul>
<p>A similar method is also used for Deep Learning. Weights of the model are sometimes randomly assigned, and sometimes, which usually result in better accuracy, they are obtained based on the pre-trained models on other tasks. That is the exact thing you will do when you remember your previous experience to use it for this new test.</p>
<h3 id="take-a-sample-test--forward-pass">Take a Sample Test — Forward Pass</h3>
<figure id="figure-photo-by-jessica-lewishttpsunsplashcomthepaintedsquareutm_sourcemediumutm_mediumreferral-on-unsplashhttpsunsplashcomutm_sourcemediumutm_mediumreferral">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/exam.jpeg" data-caption="Photo by &lt;a href=&#34;https://unsplash.com/@thepaintedsquare?utm_source=medium&amp;amp;utm_medium=referral&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;Jessica Lewis&lt;/a&gt; on &lt;a href=&#34;https://unsplash.com/?utm_source=medium&amp;amp;utm_medium=referral&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;Unsplash&lt;/a&gt;">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/exam.jpeg" alt="" >
</a>
<figcaption>
Photo by <a href="https://unsplash.com/@thepaintedsquare?utm_source=medium&amp;utm_medium=referral" target="_blank" rel="noopener">Jessica Lewis</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral" target="_blank" rel="noopener">Unsplash</a>
</figcaption>
</figure>
<p>Ok, why not take a sample test to see where you are standing? Several questions may pop into your mind when you want to start. I try to guess by answering some of mine when I started my first test. The critical factor is to extract as much information as you can from this test. So, I developed my unique method in this regard.</p>
<figure id="figure-photo-from-online-stopwatchhttpswwwonline-stopwatchcomeggtimer-countdown">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/clock.gif" data-caption="photo from &lt;a href=&#34;https://www.online-stopwatch.com/eggtimer-countdown/&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;online-stopwatch&lt;/a&gt;">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/clock.gif" alt="" >
</a>
<figcaption>
photo from <a href="https://www.online-stopwatch.com/eggtimer-countdown/" target="_blank" rel="noopener">online-stopwatch</a>
</figcaption>
</figure>
<p>Always keep the time you are devoting for each section. First, think that you are taking a real exam, so you have just an hour and not a minute more. It is better to track the time you engage with each passage. It might be a bit less than 20 minutes for the first two articles and the remaining for the last one.</p>
<p>I suppose that you are done with your 1-hour challenge. Tired? But you should not stop. Keep up the good work and finish all the parts you missed. Track the missed parts and the time you are struggling with them. After you have done all you could, this section is finished.</p>
<p>This part aims to see what your faults are and where to focus and find the priorities because It’s not efficient to learn the parts that you are already expert at them! This method is also used to train a deep model. The data is fed to the model to determine what we got first and realize what should be done next.</p>
<h3 id="analyze-your-test--back-propagation">Analyze your test — Back Propagation</h3>
<figure id="figure-photo-by-marten-newhallhttpsunsplashcomlaughayetteutm_sourcemediumutm_mediumreferral-on-unsplashhttpsunsplashcom">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/magnifier.jpeg" data-caption="Photo by &lt;a href=&#34;https://unsplash.com/@laughayette?utm_source=medium&amp;amp;utm_medium=referral&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;Marten Newhall&lt;/a&gt; on &lt;a href=&#34;https://unsplash.com/&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;Unsplash&lt;/a&gt;">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/magnifier.jpeg" alt="" >
</a>
<figcaption>
Photo by <a href="https://unsplash.com/@laughayette?utm_source=medium&amp;utm_medium=referral" target="_blank" rel="noopener">Marten Newhall</a> on <a href="https://unsplash.com/" target="_blank" rel="noopener">Unsplash</a>
</figcaption>
</figure>
<p>Up to now, you have taken your first test, and you should now focus on interpreting it. Use the answer key to find out your incorrect answers, find their cause, and try to convince yourself that the correct answer is not what you thought. After these analysis phases for each question, you can use <a href="https://www.readingielts.com/" target="_blank" rel="noopener">this</a> website to see the explained answer.</p>
<p>When you finished, give yourself two scores based on the limited time and another for the unlimited. You can also use <a href="http://ielts.calculator.free.fr/" target="_blank" rel="noopener">this</a> website to convert your raw score to see what band score you probably get.</p>
<p>Now, draw two tables. The tables might varies according to the way that you are convenient with. I explain mine. One is for each test which tells you how many mistake answers you had for each passage, and the second one is question type based table. It should give you a complete overview of how you performed based on each question type. Try to include both limited and unlimited scores in them, and after that, you are done with this phase!</p>
<figure id="figure-passage-based-analysis-table-example">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/res1.png" data-caption="Passage based analysis table example">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/res1.png" alt="" >
</a>
<figcaption>
Passage based analysis table example
</figcaption>
</figure>
<figure id="figure-question-type-based-analysis-table-example">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/res2.png" data-caption="Question type based analysis table example">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/res2.png" alt="" >
</a>
<figcaption>
Question type based analysis table example
</figcaption>
</figure>
<figure id="figure-photo-from-guru99httpswwwguru99combackpropogation-neural-networkhtml">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/back_propagation.png" data-caption="Photo from &lt;a href=&#34;https://www.guru99.com/backpropogation-neural-network.html&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;guru99&lt;/a&gt;">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/back_propagation.png" alt="" >
</a>
<figcaption>
Photo from <a href="https://www.guru99.com/backpropogation-neural-network.html" target="_blank" rel="noopener">guru99</a>
</figcaption>
</figure>
<p>A similar method that is used in training deep models is called backpropagation. After feeding the data into the model, it should find the errors in each step to know which weight in which direction must be optimized. That is the exact trick we are using in IELTS, seeing the problem (question type and passage), and how much they should be optimized to reach your goal.</p>
<h3 id="boost-your-skills--optimization">Boost your skills — Optimization</h3>
<figure id="figure-photo-by-frank-buschhttpsunsplashcomfrankbuschutm_sourcemediumutm_mediumreferral-on-unsplashhttpsunsplashcomutm_sourcemediumutm_mediumreferral">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/analyze.jpeg" data-caption="Photo by &lt;a href=&#34;https://unsplash.com/@frankbusch?utm_source=medium&amp;amp;utm_medium=referral&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;Frank Busch&lt;/a&gt; on &lt;a href=&#34;https://unsplash.com/?utm_source=medium&amp;amp;utm_medium=referral&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;Unsplash&lt;/a&gt;">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/analyze.jpeg" alt="" >
</a>
<figcaption>
Photo by <a href="https://unsplash.com/@frankbusch?utm_source=medium&amp;utm_medium=referral" target="_blank" rel="noopener">Frank Busch</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral" target="_blank" rel="noopener">Unsplash</a>
</figcaption>
</figure>
<p>This is the final step. Now you are totally aware of what is lacking and what is your strengths. Prioritize your “to be improved” components based on the distance between what you achieved and what you expected. If you can almost receive a full mark in a test with a reasonable extra time and lose much in the standard time frame, you should focus on speed. If you are slow in a specific question type, the more scrutinized decision may be focusing on some training speed tests on that particular question type and revise your answering method for that part. Generate some questions that you should look for. Be goal-oriented and avoid roaming around aimlessly.</p>
<p>Now you should look for answers. Hopefully, you are provided with a host of options, so it will not be a concern. <a href="https://www.youtube.com/" target="_blank" rel="noopener">YouTube</a> is one of the bests. Just search your produced questions (or only the keywords) and pick some guides to watch. You will learn many quick and efficient solutions from experts for free! You can also use the mentioned sources in the first stage or ask your friends or teachers about it. Try the answers in your next tests and re-evaluate yourself. If they worked, keep going and do not change it anymore and focus on another part of your new analysis table, and if they do not, try other solutions. Some optimum solutions always exist for everyone, so keep up working hard! The remaining is up to your work and how many times you repeat this 4 steps cycle.</p>
<p>Regarding the relevance of this stage to optimization, have a look at the following image:</p>
<figure id="figure-photo-from-a-amini-et-al-spatial-uncertainty-sampling-for-end-to-end-controlhttpsarxivorgpdf180504829pdf-neurips-bayesian-deep-learning-2018">
<a data-fancybox="" href="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/curve.jpeg" data-caption="Photo from: A. Amini et al. &lt;a href=&#34;https://arxiv.org/pdf/1805.04829.pdf&#34; target=&#34;_blank&#34; rel=&#34;noopener&#34;&gt;“Spatial Uncertainty Sampling for End-to-End Control”&lt;/a&gt;. NeurIPS Bayesian Deep Learning 2018">
<img src="https://raw.githubusercontent.com/amirmohammadkz/starter-academic/master/static/media/curve.jpeg" alt="" >
</a>
<figcaption>
Photo from: A. Amini et al. <a href="https://arxiv.org/pdf/1805.04829.pdf" target="_blank" rel="noopener">“Spatial Uncertainty Sampling for End-to-End Control”</a>. NeurIPS Bayesian Deep Learning 2018
</figcaption>
</figure>
<p>The goal is to reduce the error (the difference between your current score and optimum score), and the task is finding how, how much, and on which direction you should optimize. You can assume each question type as an independent dimension, and the method you choose is the direction you want to go to. The prioritization step I explained before is that focusing on some dimensions and optimizing them will not affect your result much, so focus on the ones with more space for improvements and do not waste your resources ( time and energy)!</p>
<p>These steps can be divided into more personalized stages, but I avoided telling them to keep it general. Personally, I tried to leverage some deeper connections based on my NLP experience, such as Masked Language Models, to get a better insight, and I am sure that you can do so. Build your path and break a leg!</p>
<p>If you have more sources, recommendations, and comments, I appreciate if you reach out to me.</p></description></item><item><title>personality detection using bagged SVM over BERT word embedding ensembles</title><link>https://amirmohammadkz.github.io/publication/bb-svm/</link><pubDate>Mon, 13 Jul 2020 00:00:00 +0000</pubDate><guid>https://amirmohammadkz.github.io/publication/bb-svm/</guid><description/></item></channel></rss>