Translation of the book by Andrew Eun "The Passion for Machine Learning" Chapters 1 - 14 &lt;{short}&gt; &lt;{full}&gt; <div class="post__text post__text-html js-mediator-article"> <p> Some time ago, in my Facebook feed, a link to Andrew Ng's book "

<{full}>
Some time ago, in my Facebook feed, a link to Andrew Ng's book "Machine Learning Yearning" popped up, which can be translated as "Passion for machine learning" or "Thirst for machine learning."
 
Translation of the book by Andrew Eun "The Passion for Machine Learning" Chapters 1 - 14     Some time ago, in my Facebook feed, a link to Andrew Ng's book "Machine Learning Yearning" popped up, which can be translated as "Passion for machine learning" or "Thirst for machine learning."     
      
  People who are interested in computer training or working in this field should not be represented by Andrew. For the uninitiated, it is enough to say that he is a world-magnitude star in the field of artificial intelligence. Scientist, engineer, entrepreneur, one of the founders of  Coursera  . The author of an excellent course on  introduction to machine learning  and the courses that make up the specialization "Deep Learning" (     Some time ago, in my Facebook feed, a link to Andrew Ng's book "Machine Learning Yearning" popped up, which can be translated as "Passion for machine learning" or "Thirst for machine learning."     
      
  People who are interested in computer training or working in this field should not be represented by Andrew. For the uninitiated, it is enough to say that he is a world-magnitude star in the field of artificial intelligence. Scientist, engineer, entrepreneur, one of the founders of  Coursera  . The author of an excellent course on  introduction to machine learning  and the courses that make up the specialization "Deep Learning" ( "alt =" image "/>" />
 
People who are interested in computer training or working in this field should not be represented by Andrew. For the uninitiated, it is enough to say that he is a world-magnitude star in the field of artificial intelligence. Scientist, engineer, entrepreneur, one of the founders of
Coursera
. The author of an excellent course on
introduction to machine learning
and the courses that make ...
+ 0 -

Searching for an object on the image using the perceptive hash

Searching for an object on the image using the perceptive hash
 
In those days, when I still believed that programming for myself at lunchtime or after work, you can create your startup, I had one project. And the project required such an algorithm for finding an object on the image, so that it could be quickly trained on a new object, and that it did not consume a lot of computing resources. After reading the articles about the perceptive hash ( times
? article and
two
), I decided, why not use it to limit the number of image areas to be examined? And he started build your bike, instead of using the signs of Haar. Use the perceptual hash as a filter ...
+ 0 -

Truth and lie of face recognition systems

Perhaps there is no other technology today, around which there would be so many myths, lies and incompetence. The journalists who tell about the technology are lying, politicians who talk about successful implementation lie, most tech vendors lie. Every month, I see the consequences of how people try to introduce face recognition into systems that can not work with it.
 
Truth and lie of face recognition systems  
The topic of this article long ago piqued, but it was all somehow lazy to write it. There is a lot of text that I've repeated to different people twenty times already. But, after reading the next pack of trash still decided it was time...
+ 0 -

The Amazon Rekognition face recognition system accepted 28 US congressmen for criminals

The Amazon Rekognition face recognition system accepted 28 US congressmen for criminals  
 
The American Civil Liberties Union (ACLU) continues the campaign against the use of face detection systems by federal agencies and the police. ACLU insists that the quality of these systems is too low for real use. As a result, there will be many false positives, because of which innocent people will suffer. Trying to convince the US Congress to ban the use of these systems, human rights activists took a bold, but effective action: they drove through the face recognition system Amazon Rekognition all American congressmen. The result was a bit ...
+ 0 -

AI, practical course. Modern architecture of deep neural networks for the classification of images

AI, practical course. Modern architecture of deep neural networks for the classification of images  
 
In the previous article, Overview of neural networks for the classification of images , we got acquainted with the basic basic concepts of convolutional neural networks, as well as the underlying ideas. In this article, we will examine several architectures of deep neural networks with a large processing power - such as AlexNet, ZFNet, VGG, GoogLeNet and ResNet - and we will summarize the main advantages of each of these architectures. The structure of the article is based on the blog entry ...
+ 0 -

A simple image editor on VueJS

Recently I got the opportunity to write a service for an online store, which would help to issue an order for printing their photos.
 
 
The service assumed the existence of a "simple" image editor, the creation of which I would like to share. And all because among the abundance of various plug-ins I have not found a suitable functional, in addition, the nuances of CSS transformations, suddenly became a very nontrivial task for me.
 
 
A simple image editor on VueJS
 
here .
 
 
We need two components:
 
 
Edit - will contain the basic ...
+ 0 -

Russian and Ukrainian teams took over Europeans at the European final of the Intel InnovationFPGA

Russian and Ukrainian teams took over Europeans at the European final of the Intel InnovationFPGAcompetition.Gold went to Russia, silver was shared by Russia and Italy, bronze went to Ukraine. These are the results of the European finale of the prestigious InnovateFPGA competition under the aegis of Intel. The winners will go to California, where they will meet with finalists from America and Asia. I hope, now it will not be necessary to explain on Habr, why Verilog and FPGA /FPGA are strategically important, despite the fact that "there are more vacancies for Java".  Students who are now making projects on FPGAs in a few years will make mass chips for self-controlled cars, neural network accelerators...
+ 0 -

AI, practical course. Overview of neural networks for the classification of images

This article provides an available theoretical overview of convolutional neural networks (Convolutional Neural Network, CNN) and explains their application to the task of classifying images.
 
 
AI, practical course. Overview of neural networks for the classification of images  
"Application of gradient learning to the task of document recognition" (Gradient-based Learning Applied to Document Recognition) by I. LeCun and other authors.
 
 
CNN kills two birds with one stone: there is no need to pre-define the filters, and the learning process becomes seamless. The typical CNN architecture consists of the following parts:
 
 
Convolutional layers
 
Layers of the subsample
 
Dense ...
+ 0 -

Machine vision for retail. How to read the price tags in the store

Machine vision for retail. How to read the price tags in the storeMachine vision is a very urgent topic these days. To solve the task of recognizing store price tags using neural networks, we chose the TensorFlow framework.
 
 
The article will deal specifically with how to localize and identify several objects on one shop price tag, and also to recognize its contents. A similar task of recognizing the price tags of IKEA was already decided on Habr using the classic image processing tools available in the OpenCV library.
 
 
Separately, I would like to note that the solution can work both on the SAP HANA platform in conjunction with Tensorflow Serving...
+ 0 -

AI, practical course. Preprocessing and addition of data with images

AI, practical course. Preprocessing and addition of data with imagesPreprocessing is a general term for all manipulations performed with data before the transfer of their model, including centering, normalization, shifting, rotation, pruning, and so on. Typically, preprocessing is required in two cases.
 
 
 
Data cleaning . Suppose that there are some artifacts on the images. To facilitate learning the model, the artifacts must be removed during the preprocessing phase.
 
Addition of data . Sometimes small data sets are not enough for a quality deep learning model. The data supplement approach is very useful in solving this problem. This is the process ...
+ 0 -