<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>zachmoshe.com</title>
    <description>All about data, data mining and new technologies experiments</description>
    <link>http://zachmoshe.com/</link>
    <atom:link href="http://zachmoshe.com/feed.xml" rel="self" type="application/rss+xml" />
    <pubDate>Fri, 06 May 2022 11:59:32 +0300</pubDate>
    <lastBuildDate>Fri, 06 May 2022 11:59:32 +0300</lastBuildDate>
    <generator>Jekyll v3.9.2</generator>
    
      <item>
        <title>SumOfGaussians layer with Keras 2.0</title>
        <description>How to implement a sum of gaussians model as a layer in Keras using the new TF 2.0 API.
</description>
        <pubDate>Thu, 01 Aug 2019 00:00:00 +0300</pubDate>
        <link>http://zachmoshe.com/2019/08/01/sum-of-gaussians-layer-with-keras-2.0.html</link>
        <guid isPermaLink="true">http://zachmoshe.com/2019/08/01/sum-of-gaussians-layer-with-keras-2.0.html</guid>
        
        <category>keras2.0</category>
        
        <category>tensorflow2.0</category>
        
        <category>gaussians</category>
        
        
      </item>
    
      <item>
        <title>Use Keras Pretrained Models With Tensorflow</title>
        <description>Some useful tips for using Keras pretrained models (keras.applications) in your own Tensorflow graphs.
</description>
        <pubDate>Sat, 11 Nov 2017 00:00:00 +0200</pubDate>
        <link>http://zachmoshe.com/2017/11/11/use-keras-models-with-tf.html</link>
        <guid isPermaLink="true">http://zachmoshe.com/2017/11/11/use-keras-models-with-tf.html</guid>
        
        <category>keras</category>
        
        <category>tensorflow</category>
        
        <category>vgg</category>
        
        
      </item>
    
      <item>
        <title>Simpsons Detector</title>
        <description>A summary of my CS231n course project - A Convolutional Neural Network that detects the main four Simpsons characters.
</description>
        <pubDate>Wed, 03 May 2017 00:00:00 +0300</pubDate>
        <link>http://zachmoshe.com/2017/05/03/simpsons-detector.html</link>
        <guid isPermaLink="true">http://zachmoshe.com/2017/05/03/simpsons-detector.html</guid>
        
        <category>simpsons</category>
        
        <category>deep-learning</category>
        
        <category>python</category>
        
        <category>transfer-learning</category>
        
        <category>training-set-augmentation</category>
        
        
      </item>
    
      <item>
        <title>Pickling Keras Models</title>
        <description>A short post on how to serialize Keras Model objects with python's Pickle library
</description>
        <pubDate>Mon, 03 Apr 2017 00:00:00 +0300</pubDate>
        <link>http://zachmoshe.com/2017/04/03/pickling-keras-models.html</link>
        <guid isPermaLink="true">http://zachmoshe.com/2017/04/03/pickling-keras-models.html</guid>
        
        <category>python</category>
        
        <category>deep-learning</category>
        
        <category>keras</category>
        
        <category>pickle</category>
        
        
      </item>
    
      <item>
        <title>Efficient Range-Joins With Spark 2.0</title>
        <description>How to efficiently join two Spark DataFrames on a range condition? The naive approach will end up with a full Cartesian Product and a filter, and while the generic solution to the problem is not very easy, a very popular use-case is to have join records based on timestamp difference (e.g. join every event to all measurements that were taken in the hour before its timestamp). This allows a quick win with a more efficient way to do the join as demonstrated in the post.
</description>
        <pubDate>Mon, 26 Sep 2016 00:00:00 +0300</pubDate>
        <link>http://zachmoshe.com/2016/09/26/efficient-range-joins-with-spark.html</link>
        <guid isPermaLink="true">http://zachmoshe.com/2016/09/26/efficient-range-joins-with-spark.html</guid>
        
        <category>spark</category>
        
        <category>spark-sql</category>
        
        <category>range-join</category>
        
        
      </item>
    
      <item>
        <title>Cheapest Single-Node Docker Cluster</title>
        <description>How to launch the cheapest single-node Docker cluster? This post will compare fully managed docker clusters from Google (Google Container Engine) and Amazon (AWS Container Service). Since both products are similar in capabilities (or at least, cover my basic needs), the effort will be to try and get the cheapest node possible. That does the work for me mainly as a playground or for very tiny jobs/APIs.
</description>
        <pubDate>Thu, 28 Apr 2016 00:00:00 +0300</pubDate>
        <link>http://zachmoshe.com/2016/04/28/cheapest-single-node-docker-cluster.html</link>
        <guid isPermaLink="true">http://zachmoshe.com/2016/04/28/cheapest-single-node-docker-cluster.html</guid>
        
        <category>ecs</category>
        
        <category>docker</category>
        
        <category>cluster</category>
        
        
      </item>
    
      <item>
        <title>Generating Random Polygons</title>
        <description>How to generate many non-overlapping random polygons on a plane? While doing so, we'll also look on random-grids (i.e. grids in which every boundary between two cells is probabilistic) and estimate the average block size by the size of the grid and the probability.
</description>
        <pubDate>Wed, 11 Nov 2015 00:00:00 +0200</pubDate>
        <link>http://zachmoshe.com/2015/11/11/random-polygons.html</link>
        <guid isPermaLink="true">http://zachmoshe.com/2015/11/11/random-polygons.html</guid>
        
        <category>polygons</category>
        
        <category>random-grid</category>
        
        <category>connected-components</category>
        
        
      </item>
    
      <item>
        <title>The Griddlers Ninja new serverless AWS architecture</title>
        <description>How to design and build a serverless architecture website, using various AWS products to cut down costs and avoid monthly payments. The webapp is based on AngularJS for the frontend (backed by S3 to host it), SNS and SQS to manage messages and work queues, AWS Lambda to manage the spot instances (workers) and my original python code to actually solve the boards.
</description>
        <pubDate>Wed, 29 Jul 2015 00:00:00 +0300</pubDate>
        <link>http://zachmoshe.com/2015/07/29/griddlers-ninja-serverless-architecture.html</link>
        <guid isPermaLink="true">http://zachmoshe.com/2015/07/29/griddlers-ninja-serverless-architecture.html</guid>
        
        <category>architecture</category>
        
        <category>serverless</category>
        
        <category>griddlers-ninja</category>
        
        
      </item>
    
      <item>
        <title>Looking for the state of TLV - IL elections part #1</title>
        <description>Trying to look at Israel's latest elections results and form homogeneous clusters of ballots. Will we get the &quot;state of TLV&quot; in the data as well?
</description>
        <pubDate>Tue, 24 Mar 2015 00:00:00 +0200</pubDate>
        <link>http://zachmoshe.com/2015/03/24/israel-elections-part-1.html</link>
        <guid isPermaLink="true">http://zachmoshe.com/2015/03/24/israel-elections-part-1.html</guid>
        
        <category>elections</category>
        
        <category>clustering</category>
        
        
      </item>
    
      <item>
        <title>Use Ruby Gems With Hadoop Streaming</title>
        <description>How to use ruby gems in a script that goes to Hadoop Streaming? Dealing with dependency management, packaging and deploying the archive to the cloud.
</description>
        <pubDate>Mon, 23 Feb 2015 00:00:00 +0200</pubDate>
        <link>http://zachmoshe.com/2015/02/23/use-ruby-gems-with-hadoop-streaming.html</link>
        <guid isPermaLink="true">http://zachmoshe.com/2015/02/23/use-ruby-gems-with-hadoop-streaming.html</guid>
        
        <category>hadoop</category>
        
        <category>ruby</category>
        
        
      </item>
    
  </channel>
</rss>
