Upload GAN post sketch

_config.yml

@@ -17,12 +17,15 @@ paginate:            5
 markdown:            kramdown
 highlighter:         rouge
 permalink:           none
-plugins:             [jekyll-paginate, jekyll-sitemap, jekyll-feed, jekyll-seo-tag]
+plugins:             [jekyll-paginate, jekyll-sitemap, jekyll-feed, jekyll-seo-tag, jekyll-crosspost-to-medium]
 
 # Customise atom feed settings (this is where Jekyll-Feed gets configuration information)
-title:		           "Lagrange"
-description:	       "A minimalist Jekyll theme"
-author:              "Paul Le"
+title:		           "DTRANSPOSED"
+description:	       "data science, AI, technology"
+author:              "Damian Bogunowicz"
+
+# Analytics
+analytics-id:        "UA-131652597-1"
 
 # RSS 2.0 can be used instead of Atom by uncommenting following two lines
 #feed:

_data/authors.yml

@@ -1,3 +1,2 @@
 primary:
-  name: 'Paul Le'
-  twitter:  paululele
+  name: 'Damian Bogunowicz'

_data/settings.yml

@@ -1,24 +1,28 @@
-title:  Lagrange
-tagline:  a minimalist Jekyll theme
+title: DTRANSPOSED
+tagline:  data science, AI, technology
 
 # Authors documented under _data
 
 # Change to true for Disqus comments
 disqus:
-  comments: false
-  disqus_shortname: 'https-lenpaul-github-io-lagrange'
+  comments: true
+  disqus_shortname: 'https-dtransposed-github-io'
 
 google-ID: 'UA-80213386-1'
 
 menu:
+- {name: 'Blog', path: 'menu', url: 'blog.html'}
 - {name: 'About', path: 'menu', url: 'about.html'}
-- {name: 'Writing', path: 'menu', url: 'writing.html'}
-- {name: 'Contact', path: 'menu', url: 'contact.html'}
+- {name: 'Projects', path: 'menu', url: 'projects.html'}
 
 # Available social icons are powered by Font Awesome, so you can use any icon that they offer
 social:
-- {icon: 'github', link: 'https://www.github.com/lenpaul'}
-- {icon: 'twitter', link: 'https://twitter.com/paululele'}
-- {icon: 'linkedin', link: 'http://www.linkedin.com/in/lenpaul/'}
-- {icon: 'envelope', link: 'mailto:l.nguyen.paul@gmail.com'}
+- {icon: 'github', link: 'https://www.github.com/dtransposed'}
+- {icon: 'twitter', link: 'https://twitter.com/dtransposed'}
+- {icon: 'linkedin', link: 'http://www.linkedin.com/in/dbogunowicz/'}
+- {icon: 'envelope', link: 'mailto:damian.bogunowicz@gmail.com'}
 - {icon: 'rss-square', link: 'feed.xml'}
+
+pagination:
+  next_page: 'Newer'
+  previous_page: 'Older'

_drafts/Adversarial-Attack.md

@@ -0,0 +1,56 @@
+---
+layout: post
+title: "Adverserial attacks"
+author: "Damian Bogunowicz"
+categories: blog
+tags: [AI, hidden markov model, tutorial, python, programming, psychology]
+image: face-recognition.jpg
+---
+self driving cars,
+no longer need human drivers in the loop
+identify human faces airports
+security systems that can identify people present on the company ground
+adverserial attack
+
+ $$\mathcal{L}=\alpha\mathcal{L}_{in}(x,x')+\beta\mathcal{L}_{out}((f(x'),f(x))+\gamma\mathcal{L}_{s}$$
+
+The loss function in general consists of three components. Let's take a look at each of these separate losses.
+
+## Perturbation loss
+
+The goal of the adversarial attack is to take an original, benign input $$x$$ and produce a malicious image $$x'$$ by adding some noise (perturbation) $$\eta$$ to it. In other words $$x'=x+\eta$$. It would be great to fool a network while keeping the noise added to the image to the minimum. This means that one of the primary goals of the adversarial network is to minimize the difference between the original image $$x$$ and the perturbated image $$x'$$. The mathematical formulation of this condition is:
+
+$$ \underset{\eta}{\operatorname{argmin}} \mathcal{L}_{in}(x,x')$$
+
+This means that we manipulate the noise in such a way, that it minimizes the loss function, i.e. a malicious input has possibly close pixel values to the original image. 
+
+Let's take an autonomous vehicle as an example. Let's step into shoes of a villain, who wants to attack a convoy of autonomous cars.  Perturbation loss would be a crucial element of a physical attack on a network. 
+
+##
+
+The authors of the paper Adversarial Generative Nets: Neural Network Attacks on State-of-the-Art Face Recognition    In impersonation (or targeted) attacks, the adversary attempts to be misclassified as a specific other person.
+-	In dodging (or untargeted) attacks, the adversary attempts to be misclassified as some other person. 
+
+autonomous car sees sign as something different
+impersonate a person using physical artifact lie 
+
+
+
+
+Term               | Definition              |          
+--------------------- | :-------------------: | 
+a benign image $$x$$                | original input to the network            | 
+an adversarial image $$x'$$  | an image that has been altered in order to fool the classifier | 
+non-targeted attacks | an attack where the attacker desires to be missclassified by any means|
+targeted attack | an attack where the attacker tries to get missclassified to a specific target class |
+white box | yfftftffffghf|
+black box attack | when the attacker has the access to the model's architecture (parameter |
+physical attack/digital attack |ygfyfyyf/|
+
+
+
+ine learning models that an attacker has intentionally designed to cause the model to make a mistake; they?re like optical illusions for machines. In this post we?ll show how adversarial examples work across different mediums, and will discuss why securing systems against them can be difficult.
+
+ Explaining and Harnessing Adversarial Examples:  panda image
+
+<em>Source of the cover image: http://www.bleum.com</em>

_drafts/alphazero.md

@@ -0,0 +1,22 @@
+earning minmax
+
+<https://www.youtube.com/watch?v=fInYh90YMJU>
+
+quick alpha zero:
+
+<https://tmoer.github.io/AlphaZero/>
+
+<https://jeremykun.com/2013/10/28/optimism-in-the-face-of-uncertainty-the-ucb1-algorithm/>
+
+
+
+Sources:
+
+Creating AlphaZero:
+https://medium.com/applied-data-science/how-to-build-your-own-alphazero-ai-using-python-and-keras-7f664945c188
+
+AlphaZero Cheatsheet:
+https://medium.com/applied-data-science/alphago-zero-explained-in-one-diagram-365f5abf67e0
+
+AlphaZero Stanford:
+https://web.stanford.edu/~surag/posts/alphazero.html

_drafts/maiot.md

@@ -0,0 +1 @@
+“Every person in your company is a vector. Your progress is determined by the sum of all vectors.” — Elon Musk

_drafts/sim2real

@@ -0,0 +1,13 @@
+
+Sim-to-Real Transfer of Robotic Control with Dynamics Randomization
+
+Using Simulation and Domain Adaptation to Improve Efficiency of Deep Robotic Grasping
+Sim-to-Real via Sim-to-Sim: Data-efficient Robotic Grasping via Randomized-to-Canonical Adaptation Networks
+
+Closing the Sim-to-Real Loop: Adapting Simulation Randomization with Real World Experience
+
+Robots Learning to Move like Animals
+
+Meta-Reinforcement Learning for Robotic Industrial Insertion Tasks
+
+SOLVING RUBIK’S CUBE WITH A ROBOT HAND

_includes/google-analytics.html

@@ -1,9 +1,9 @@
+<!-- Global site tag (gtag.js) - Google Analytics -->
+<script async src="https://www.googletagmanager.com/gtag/js?id=UA-131652597-1"></script>
 <script>
-  (function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){
-  (i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o),
-  m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m)
-  })(window,document,'script','//www.google-analytics.com/analytics.js','ga');
+  window.dataLayer = window.dataLayer || [];
+  function gtag(){dataLayer.push(arguments);}
+  gtag('js', new Date());
 
-  ga('create', '{{ site.data.settings.google-ID }}', 'auto');
-  ga('send', 'pageview');
+  gtag('config', 'UA-131652597-1');
 </script>

_includes/head.html

@@ -33,5 +33,4 @@
 
   <!-- Use Jekyll SEO plugin -->
   {% seo %}
-
 </head>

_layouts/default.html

@@ -7,6 +7,7 @@
 
   <div class="container">
     {% include header.html %}
+    {% seo %}
 
     <div class="post-container">
       {{ content }}

_layouts/post.html

@@ -6,14 +6,14 @@ <h1>
   {{ page.title }}
 </h1>
 
-{% if page.image.feature %}
-  <img src="{{ site.github.url }}/assets/img/{{ page.image.feature }}">
+{% if page.image %}
+  <img src="{{ site.github.url }}/assets/img/{{ page.image }}">
 {% endif %}
 
 {{ content }}
 
 <span class="post-date">
-  Written on
+  {{ site.data.settings.post_date_prefix }}
   {% assign d = page.date | date: "%-d"  %}
   {{ page.date | date: "%B" }}
   {% case d %}
@@ -26,7 +26,7 @@ <h1>
   {% if page.author %}
     {{ page.author }}
   {% else %}
-    {{ site.data.authors.primary.name }}
+    {{ site.author }}
   {% endif %}
 </span>
 
@@ -35,5 +35,21 @@ <h1>
 {% include related-posts.html %}
 
 {% if site.data.settings.disqus.comments %}
+<div id="disqus_thread"></div>
+<script>
+var disqus_config = function () {
+this.page.url = https://dtransposed.github.io;  // Replace PAGE_URL with your page's canonical URL variable
+this.page.identifier = PAGE_IDENTIFIER; // Replace PAGE_IDENTIFIER with your page's unique identifier variable
+};
+*/
+(function() { // DON'T EDIT BELOW THIS LINE
+var d = document, s = d.createElement('script');
+s.src = 'https://https-dtransposed-github-io.disqus.com/embed.js';
+s.setAttribute('data-timestamp', +new Date());
+(d.head || d.body).appendChild(s);
+})();
+</script>
+<noscript>Please enable JavaScript to view the <a href="https://disqus.com/?ref_noscript">comments powered by Disqus.</a></noscript>
+
   {% include disqus.html %}
 {% endif %}