 
  


         ()   .  :    ;  ,   ;       REST API;   :  , ,  , ,   ,  , , ;       ,   ;     ML-   C#!






1   AI?


. ,       (   )      (   ),   2-  -.



,    AI, ML,  ,  BigData, AR/VR,     .       ,  ,    .



  .        PRISM  MSQRD.        ,       $   .



    ,      .      (, , , ),    (, , ),  (, , , , ?),  .     ?



      ,     API  .  , , , , -  .



  , , , :

1)     (       ,  ).

2)   -   . -    .   -   .    ,        .                 ,          (  )   .     ,     .



    ,    ,        .  ,             VK/OK.       .         .    :

1) ,   .

2)  ,        .

3)  ,      .

4)  ,     .     (  ,     ).



,      .  5 ,  4  $/,    ~10 .  200  $.     3-5   0.6-1  $.



   data science      :

1)       , , , 5500 /, 250 .

2) Data science  SkillFactory, 120  , 12 .

3) Data scientist  , 200  , 8 .

4)      , , , , 35 .

5) Introduction to Machine Learning  Google, , 15 .

https://developers.google.com/machine-learning/crash-course (https://developers.google.com/machine-learning/crash-course)

6)  ?



 4     #5  ,   2  . -,   -,       .



 ?



















  (https://ru.m.wikipedia.org/wiki/%D0%A4%D0%B0%D0%B9%D0%BB:AI-ML-DL.png)










  (https://commons.wikimedia.org/wiki/File:Artificial_neural_network_image_recognition.png)










  (https://en.wikipedia.org/wiki/File:Ai_integration.jpg)

, . https://en.wikipedia.org/wiki/Applications_of_artificial_intelligence (https://en.wikipedia.org/wiki/Applications_of_artificial_intelligence)










  (https://ru.wikipedia.org/wiki/%D0%A4%D0%B0%D0%B9%D0%BB:Gartner_Hype_Cycle_for_Emerging_Technologies.gif)




2  AI-


    AI (.  ).     , ,    AI-,     .   AI- , YC, 500 Startups  TechStars.



..       ,      YC:

1)   .    Summer 2019  175  27 . , , , ,  .  ,   .     - .

2)   7%  $120  175         .    3-5    . ,  YC      $$$     .         ,  ,   ,      -     .

3) ,      . 11 vs 175  ( vs YC)   .           ,           .  ,    149-   ,     :-(   ?

4)    YC   ,  -,    .       -, , , , -, .

5)  YC   .     ,  , , .       175       .

6) AI   .    AI    .       , -   ( ,   -  1-10  ).

7)    YC  ,        .    ,       .

8) ,   !      ,   , VR  ,    ,     . , ,  ,  !         :-)



, AI-  .



1)  #19, 3  2019 (11 ):

https://www.iidf.ru/media/articles/accelerator/19-nabor-akseleratora/ (https://www.iidf.ru/media/articles/accelerator/19-nabor-akseleratora/)



Leveli.ng ()  AI-        .   ,         :  ,  ,    .



2)  #18, 1  2019 (19 ):

https://www.iidf.ru/media/articles/accelerator/v-18-nabor-akseleratora/ (https://www.iidf.ru/media/articles/accelerator/v-18-nabor-akseleratora/)



Clover Group ()               ,     .



FoTail (--)  -       .  IT-      ,   big data  machine learning,      ,        .



Music Scan ()  ,   Big Data,      -,               .



 ()   ,     .         ,             .



3)YC Summer 2019 Batch (175 companies):

https://blog.ycombinator.com/yc-summer-2019-batch-stats/ (https://blog.ycombinator.com/yc-summer-2019-batch-stats/)

https://techcrunch.com/2019/08/19/all-84-startups-from-y-combinators-s19-demo-day-1/ (https://techcrunch.com/2019/08/19/all-84-startups-from-y-combinators-s19-demo-day-1/)

https://techcrunch.com/2019/08/20/here-are-the-82-startups (https://techcrunch.com/2019/08/20/here-are-the-82-startups-that-launched-on-day-2-of-ycs-s19-demo-days/)



Intersect Labs: Intersect Labs is building CoreML for enterprise, letting its customers easily build machine learning models to help make sense of their historical data and deliver insights without having to hire data scientists. The monthly subscription is aiming to deliver a product that doesnt require much technical knowledge. If you can use a spreadsheet, you can use Intersect Labs.



Traces: As privacy-conscious consumers speak up against the proliferation of facial recognition tech, theres still a clear need for a product that enables smart camera tracking for customers. Traces is building computer vision tracking tech that relies on cues other than facial structure like clothing and size to help customers integrate less invasive tracking tech. It was built by former Ring engineers.



Soteris: Soteris is a startup building machine learning software for insurance pricing. Within si months of their pilot, they already have two insurers under contract, giving them $500K in guaranteed annual revenue.



Well Principled: This is an AI-driven management consultant that says it wants to replace MBAs with software. Companies spend $200 billion on management consultants every year. Well Principled wants to replace that epensive and cumbersome system with its tech that has culled growth and revenue learnings from academic research and turned it into enterprise software. The company wants to eliminate the need for outside consultants by integrating its software into the daily operations of businesses as they launch new products. Well Principled is advised and invested in by early Palantir leaders, and claims $840,000 ARR from its first Fortune 200 customer.



Dashblock: Dashbloack creates APIs from any web page using machine learning. Drop in a URL, select the data you want from a page, and it will figure out how to automatically etract it and provide it via API. It has have more than 1,500 users since launching two weeks ago.



EARTH AI: This full stack AI-powered mining eploration company built a technology to predict the location of un-mined rare metals. EARTH AIs mission is to improve the efficiency of mineral eploration to provide enough metals and minerals for current and future generations. The company predicts where metals may eist, actually mines the ore and then sells it. The team credits themselves with discovering the worlds first AI-predicted mineral deposit, and says it has also secured the rights to $18 billion worth of ore.



Holy Grail: Holy Grail says it has built a cheaper and faster way to manufacture batteries. The company is using AI to find the net generation of batteries at what it claims is 1,000 faster and hundreds of million dollars cheaper than traditional R&D processes. Holy Grails software designs batteries and predicts their performance  then manufactures them using a robot it built. Traditional R&D relies on trial and error and spreadsheets, and this company thinks it can harness AI to do something good for the world while also making money.



Zenith: This company is building a new virtual world that blends AI, VR and its backend tech to immerse users in new lives online. Zenith, which raised $120,000 on Kickstarter in one week, is the first cross platform world to eist on VR desktop and console. Essentially every screen you own is a window into their world. The company plans to monetize by taking cuts of every item bought or sold on their platform, like property and clothing. The founders have worked at Google and Unity, and co-produced with Oculus.



Lofty AI: Lofty AI is building what they claim to be the first reliable method for tracking neighborhood demand to help real estate investors make more informed investment decisions. Lofty AI recommends properties to investors and if the investors decide to purchase, they enter into a contract that gives them 20% of the profit. However, if the value of a property goes down, Lofty says they will cover all of the investors losses.



Treble.ai: This is a customer support platform that lets companies get feedback from users through SMS and WhatsApp. The company describes itself as similar to Qualtrics and Zendesk, but with one big difference: Qualtrics and Zendesk were built for desktop web and email. Treble is built for mobile-first, chat-based communication. Treble says there are 100,000 companies that serve their users through mobile apps, and it wants to be the startup that manages their customer support. The startup scored Colombian logistics unicorn Rappi as their largest customer, and is seeing $16,000 in MRR.



Symple: The team at Symple is developing an AI-based doctor that can diagnose you using your smartphone. The startup says theyve signed up 15 doctors in the first few weeks, with a goal of epanding into a $2.6 billion market. Heres how it works: First, you tell Symple how youre feeling, then, the companys machine learning algorithm gauges your condition and provides a detailed initial diagnosis, which is then stored and saved.



Percept.AI: This startup is creating an AI support agent that immediately resolves common customer support tickets. Other solutions can take over three weeks of onboarding, quality is often insufficient and the AIs only end up resolving between 10-30% of tickets. Percept.AI says their tech could work to identify 1.2 billion support tickets that go outstanding. They say they can immediately resolve up to 50% of tickets without human intervention, what it describes as an eciting $22 billion market.



4)500 Startups, Batch #25, JULY 22, 2019:

https://500.co/meet-the-startups-of-batch-25/ (https://500.co/meet-the-startups-of-batch-25/)

https://500.co/latest-ai-applications-from-500s-batch-25/ (https://500.co/latest-ai-applications-from-500s-batch-25/)

https://500.co/startups/ (https://500.co/startups/)



Chemtech: An AI-product for manufacturing plant automatization

Curie: A camera-based AI shopping assistant

EINO: An AI platform that produces meaningful predictive and historical insights on localized population movement and their intention in urban areas for enterprise business users

Hearte: Helping companies quickly build AI products and features

InnerTrends: A data science service for SaaS that uncovers deep insights in customer onboarding, retention, and engagement without the need of data scientists

LucidAct Health: An AI assistant for nurses and case managers to help them know what to do faster and eliminate errors

RestAR: 3D capturing and product visualization for e-commerce using AI with any mobile device

Visionful: Connecting smart cities and autonomous vehicles leveraging AI and computer vision to provide full automation for parking and traffic monitoring



5)TechStars,  AI/ML  Analytics:

https://www.techstars.com/companies-in-program/ (https://www.techstars.com/companies-in-program/)



Asgard.ai: We think any sales intelligence solutions with a broad scope might want to acquire us as they have difficulties to distinguish from one to another. They would typically buy our signal detection engine. Currently, we are able to detect 32 kinds of signals. Our goal is to improve it so it can track more specific buying signals for every new customer. And even maybe to design a product that lets the customer easily set up a very relevant signal by giving feedback/interacting on the platform.



FLOD: FLOD develops a multi-sensor based learning algorithm for structural maintenance detection and prediction.



Heuristech: Heuristech is about dealing with pipeline maintenance the right way.



Lightmass Dynamics: An application framework that easily integrates with any software that manipulates n-dimensional data for real-time simulation or visualization of physical forces.



MorphL AI: MorphL drastically reduces the amount of time required to infuse AI into your applications so you can predict user behaviors, enable personalized digital eperiences and increase product KPIs.



ReelData AI: AI for aquaculture. Delivers real time weight distributions, feeding analysis, health analysis and growth analysis.



Revelio Labs: Revelio Labs provides an in-depth view into the workforces of companies around the world. We leverage advancements in AI technologies to turn hundreds of millions of online profiles, resumes, and job postings into clear insights. Our clients include: corporate strategists, HR, VCs, and investors.



Shipright: Shipright helps software businesses track customer feedback in one organized place, so they can build outstanding products.



-    AI- , ,    (.  ).



:

1)  - AI The New Business of AI (and How Its Different From Traditional Software):

https://a16z.com (https://a16z.com/2020/02/16/the-new-business-of-ai-and-how-its-different-from-traditional-software/)

2)     ?:

https://habr.com/ru/company/mailru/blog/481256/ (https://habr.com/ru/company/mailru/blog/481256/)

 :

) AI    ,       .  ,  UX, , ,     ,   AI  .

) AI     .  , ,    10-15 (        ),       .

)   AI  (        ).   ,        . , , , , , -      ,  :grimacing:,   .   AI,        .          ,     .    AI    ,   .










  (https://commons.wikimedia.org/wiki/File:YCombinator_(2014).png)










  (https://commons.wikimedia.org/wiki/File:Early-stage-investors-startup-studios.jpeg)




3  


,      .           ()     :

https://ru.wikipedia.org/wiki/__ (https://ru.wikipedia.org/wiki/%D0%98%D1%81%D0%BA%D1%83%D1%81%D1%81%D1%82%D0%B2%D0%B5%D0%BD%D0%BD%D0%B0%D1%8F_%D0%BD%D0%B5%D0%B9%D1%80%D0%BE%D0%BD%D0%BD%D0%B0%D1%8F_%D1%81%D0%B5%D1%82%D1%8C)

https://ru.wikipedia.org/wiki/ (https://ru.wikipedia.org/wiki/%D0%9F%D0%B5%D1%80%D1%86%D0%B5%D0%BF%D1%82%D1%80%D0%BE%D0%BD)

https://ru.wikipedia.org/wiki/,_ (https://ru.wikipedia.org/wiki/%D0%A0%D0%BE%D0%B7%D0%B5%D0%BD%D0%B1%D0%BB%D0%B0%D1%82%D1%82,_%D0%A4%D1%80%D1%8D%D0%BD%D0%BA)

https://ru.wikipedia.org/wiki/_ (https://ru.wikipedia.org/wiki/%D0%98%D1%81%D0%BA%D1%83%D1%81%D1%81%D1%82%D0%B2%D0%B5%D0%BD%D0%BD%D1%8B%D0%B9_%D0%B8%D0%BD%D1%82%D0%B5%D0%BB%D0%BB%D0%B5%D0%BA%D1%82)

https://ru.wikipedia.org/wiki/__ (https://ru.wikipedia.org/wiki/%D0%97%D0%B8%D0%BC%D0%B0_%D0%B8%D1%81%D0%BA%D1%83%D1%81%D1%81%D1%82%D0%B2%D0%B5%D0%BD%D0%BD%D0%BE%D0%B3%D0%BE_%D0%B8%D0%BD%D1%82%D0%B5%D0%BB%D0%BB%D0%B5%D0%BA%D1%82%D0%B0)



   ,      " ,       :        ,   ". ,     !      , ,         . , ,    .



 ,     .



   100*100 .  ,   0..9   .



 1-  ( S)     100*100.     0  1    ,         .



 2-  ( A)     N*100*100.  N   ,     .    N=10, ..     0..9, ..  10  / .    [N, i, j]       ,            0..9.



  Xij   ?    100*100  -   0..9.       1  Xij,   1.



..   1, ,   ,       .       .



 ,  3  ( R)      N.               0..9. ..    ,   ,    R[n]    2  A[n, i, j].



       0..9   R[n],        .  ,      , ,   "8"     85%,   "3"  60%   .     ,                  >80%.    ,      .



    . ,    3-   .   , ..     A-.



      A-. ,  A-   , ,    .  2 A-  1  ,         .



    , 2-   ,  3 ,     .  2-    1. ,      the      3    2  "th",  3    2- ,  3-  "e".   .



    "       ",    :  ,   ,       .



    :

1)         .

2)     , ..        , ,   ,    ,   .

3)         ,    .

4)                 ,  ,     ,   ,  ,    .

5)         ( ,  ),    .

6)           -  ,     .

7)              , ,   ,  ,    .

7)    , -,    2   :

)       , .. ,     1  2,   .

)      ,    , ..         ,     ..



         :

1)    ;

2)    ;

3)   ;

4)    ;

5)    ;

6)  ;

7)   ;

8)  ,  ;

9)      .



,     .              . ,  -      . ,    .    .      ,   .       ,         2   .



   ,  , , ..     -.      -.  ,       !    ,        , ..          .



        ,     ,      , ..    . ..      .  ,            .      .    -, ,      .



      .          ,    100% .       , ..         .                   ?



 ,     .  " ".  ,      (-  ),  ,         .       1969    ,     ( )          .            (    ),  ,    ,  Lisp- (     )  .



,     .  .          /      1  2-  .  .      ,     .  ,      ,    ,              ,        .   .



  43             1971 . ,  -                      .        .    ,    .



                   .             . ,        Pi,          . ,          .



:

1)    : https://vas3k.ru/blog/machine_learning/ (https://vas3k.ru/blog/machine_learning/)

2)   : https://ods.ai (https://ods.ai)










  (https://ru.wikipedia.org/wiki/%D0%A4%D0%B0%D0%B9%D0%BB:Perceptron-ru.svg)




4   


    ().       (, ),      ,     ( " "),     .



,    .  ,   .   f() = 1/(1+e^-).    !       .    [-?..+?]   [0..1].



     ,    ( )      . ..   [0..1] , ,   [3..4].



 ,  -            .



    .    3 : ,   .      .       .        .       ,        2-     3-  .       2-        (  )     [0..1].



  0  1        .   ,  "  "       2- ,   .     "   ".       ,  ,  ,        .



 ,   ?     ?     ?      (   )   ?          ?           ?          ,         ?      f_error()       1n,          ?



 ,        .         ,     .



             HelloWorld,      .    ,   ,     0..9. -.



       ,     ,         .        kaggle.com    ,   datascience-.   *.csv  MNIST  60+10      2828        .



  .   "  ". La mthode de l'accumulation de probabilit (fr).     ,      .



  ,                 .       10  2828       0..9.



              0..9.       ,    .     ,    ,   .     "       8".        0..9.



   ,      ~100 . .       .



  .      ,         .    - ( ).            .



       .      57%    5    5  .



   ,    ,   99.8%   .       .



 ,    .         Google/TensorFlow, Microsoft/Azure, Amazon/AWS, /DSVM.                   .  -        .           .



   ,    ,         "-".      ,   DeepBlue (   )  AlphaGo (    )    ,          .



       /   ,   .   ,           ,     .



, .    ?       . ,   ,   . ,   ,    ,      ,         .    ,   .       ELIZA,          1996  (  1966 !).



     !   ,    .     "     ,    ".             .    ,    .



   ,       :

       .

             .

   ,  ,    ,             .

   (/, ,  ,    !   ,        ).



     ,    , ..    .



   ,     ,         .



      .   2- -.      ,   ,     ,         .         .



      .  , ,              .          .              .



, -12    .  ,    ,   .



1).   ,          .    !     ?



2) .      ,      .  !



3)     .         .      1-   8 /,             .   11.2/    (16.7   ).  ,     ?      ?



4)   .          ?



5)     .        50  -235     ,    ?           ,    ?



6)   .            ,          ?     ? H2+O2= H2O  .       ?    ,     .



7)       ,   ?  -    ?   = 0,   - .  !



8) ,         .



9)  (  ).           300  /?       ?   (    )  -?         ,     ?



10) - .     !      .     ,       ()  .    (!) ,     2    (!!!)       , ..  .



11)  . ,                .



12)  .    ?!      (       ).



    DataScience.   ,     .  data   science.



:

1)       (49 ): https://drive.google.com/file/d/1g1Owp6PLOPE6_ChbJoe8paLDviRczHPR (https://drive.google.com/file/d/1g1Owp6PLOPE6_ChbJoe8paLDviRczHPR/view?usp=sharing)

https://github.com/berlicon/SimpleNeuralNetworkMNIST (https://github.com/berlicon/SimpleNeuralNetworkMNIST)

2) MNIST database    *.csv:

https://www.kaggle.com/oddrationale/mnist-in-csv (https://www.kaggle.com/oddrationale/mnist-in-csv)

3) MNIST database  :

http://yann.lecun.com/exdb/mnist/ (http://yann.lecun.com/exdb/mnist/)

https://en.wikipedia.org/wiki/MNIST_database (https://en.wikipedia.org/wiki/MNIST_database)

4) https://en.wikipedia.org/wiki/List_of_datasets_for_machine-learning_research (https://en.wikipedia.org/wiki/List_of_datasets_for_machine-learning_research)

5) 52    :

https://habr.com/ru/company/edison/blog/480408/ (https://habr.com/ru/company/edison/blog/480408/)

6)    .  1:

https://habr.com/ru/post/312450/ (https://habr.com/ru/post/312450/)

7)    .  2:

https://habr.com/ru/post/313216/ (https://habr.com/ru/post/313216/)

8)    :

https://vas3k.ru/blog/machine_learning/ (https://vas3k.ru/blog/machine_learning/)

9) https://ru.wikipedia.org/wiki/___ (https://ru.wikipedia.org/wiki/%D0%9C%D0%B5%D1%82%D0%BE%D0%B4_%D0%BE%D0%B1%D1%80%D0%B0%D1%82%D0%BD%D0%BE%D0%B3%D0%BE_%D1%80%D0%B0%D1%81%D0%BF%D1%80%D0%BE%D1%81%D1%82%D1%80%D0%B0%D0%BD%D0%B5%D0%BD%D0%B8%D1%8F_%D0%BE%D1%88%D0%B8%D0%B1%D0%BA%D0%B8)

10) https://ru.wikipedia.org/wiki/_ (https://ru.wikipedia.org/wiki/%D0%93%D1%80%D0%B0%D0%B4%D0%B8%D0%B5%D0%BD%D1%82%D0%BD%D1%8B%D0%B9_%D1%81%D0%BF%D1%83%D1%81%D0%BA)

11) https://en.wikipedia.org/wiki/ELIZA (https://en.wikipedia.org/wiki/ELIZA)



//Program.cs

using System;

using System.Collections.Generic;

using System.Linq;

using System.IO;

using System.Tet;



namespace SimpleNeuralNetworkMNIST

{

class Program

{

const int IMAGE_SIZE = 28; //each image 28*28 piels

const int SAMPLE_COUNT = 10; //analyse 10 images  numbers 0..9

const int TRAIN_ROWS_COUNT = 5000; //first rows to train;

const int TEST_ROWS_COUNT = 5000; //other rows to test

const int INCORRECT_PENALTY = byte.MaValue * TRAIN_ROWS_COUNT; //penalty for incorrect overlap

//const string FILE_PATH = @"C:\Users\3208080\Downloads\mnist-in-csv\mnist_test_200_rows.csv";//43% 100+100

//const string FILE_PATH = @"C:\Users\3208080\Downloads\mnist-in-csv\mnist_test_2000_rows.csv";//53% 1000+1000

//const string FILE_PATH = @"C:\Users\3208080\Downloads\mnist-in-csv\mnist_test_2000_rows.csv";//56% 1900+100

//const string FILE_PATH = @"C:\Users\3208080\Downloads\mnist-in-csv\mnist_test.csv";//50% 9900+100

//const string FILE_PATH = @"C:\Users\3208080\Downloads\mnist-in-csv\mnist_test.csv";//56% 9000+1000

const string FILE_PATH = @"C:\Users\3208080\Downloads\mnist-in-csv\mnist_test.csv";//57% 5000+5000

//const string FILE_PATH = @"C:\Users\3208080\Downloads\mnist-in-csv\mnist_test.csv";//49% 1000+9000

//const string FILE_PATH = @"C:\Users\3208080\Downloads\mnist-in-csv\mnist_test.csv";//41% 100+9900



//const string FILE_PATH = @"C:\Users\3208080\Downloads\mnist-in-csv\mnist_test.csv";//55% 5000+5000 black/white



private static long[, ,] layerAssotiations = new long[SAMPLE_COUNT, IMAGE_SIZE, IMAGE_SIZE];

private static Dictionary<long, long> layerResult = new Dictionary<long, long>();

private static long correctResults = 0;



static void Main(string[] args)

{

train();

test();



Console.WriteLine("  {0}% ",

100 * correctResults / TEST_ROWS_COUNT);

}



private static void train()

{

Console.WriteLine("  ");

var inde = 1;

var rows = File.ReadAllLines(FILE_PATH).Skip(1).Take(TRAIN_ROWS_COUNT).ToList();



foreach (var row in rows)

{

Console.WriteLine(" {0}  {1}", inde++, TRAIN_ROWS_COUNT);

var values = row.Split(',');

for (int i = 1; i < values.Length; i++)

{

var value = byte.Parse(values[i]); //var value = (values[i] == "0") ? 0 : 1;

layerAssotiations[

byte.Parse(values[0]),

(i  1) / IMAGE_SIZE,

(i  1) % IMAGE_SIZE]

+= value;

}

}

}



private static void test()

{

Console.WriteLine("  ");

var inde = 1;

var rows = File.ReadAllLines(FILE_PATH).Skip(1 + TRAIN_ROWS_COUNT).Take(TEST_ROWS_COUNT).ToList();



foreach (var row in rows)

{

Console.WriteLine(" {0}  {1}", inde++, TEST_ROWS_COUNT);

clearResultLayer();



var values = row.Split(',');

for (int i = 1; i < values.Length; i++)

{

var value = byte.Parse(values[i]);

for (int j = 0; j < SAMPLE_COUNT; j++)

{

if (value > 0)

{

var weight = layerAssotiations[

j,

(i  1) / IMAGE_SIZE,

(i  1) % IMAGE_SIZE];

layerResult[j] += (weight >= 0) ? weight : -INCORRECT_PENALTY;

}

}

}



calculateStatistics(byte.Parse(values[0]));

}

}



private static void clearResultLayer()

{

layerResult = new Dictionary<long, long>();

for (int i = 0; i < SAMPLE_COUNT; i++) layerResult[i] = 0;

}



private static void calculateStatistics(byte correctNumber)

{

var proposalNumber = layerResult.OrderByDescending(p => p.Value).First().Key;

Console.WriteLine(" {0}   {1} {2}", correctNumber, proposalNumber,

proposalNumber == correctNumber ? "" : "");

if (proposalNumber == correctNumber) correctResults++;

}

}

}










  (https://commons.wikimedia.org/wiki/File:MnistExamples.png)










  (https://ru.m.wikipedia.org/wiki/%D0%A4%D0%B0%D0%B9%D0%BB:Extrema_example.svg)




5 Matt Mazur


, ,    ().       back propagation   .   .        0..9   MNIST.  ,  ,  .       .  ,        . ,  , .    10%  . , .       .



 . ,       , ,        PhD  :



1)           . ,        ,       .   ,    .



2)  ,          ,          ,       .



3)        ,          .   .  ,  3  4-   . ..        . ..  : (Tj-Oj)*Oj*(1-Oj),    (Tj-Oj)*Oj*(1-Oj)*On,  n=j-1



4)        .     Full-connected [FC] (     )         FC-.



5)     ,          (       ).



6)     ",        dE/dw". , .      .



7)           , ,           .



8) ,      ( 0..255, ,     0..1).



9) ,        . ,  . 0..1  0..0,001?



 ,  ,    2  !



           .  ,       .   . ..    ,         -  (  )    .



   ,           ,         .      .  - .  ,     . ,  .



 ,      ,   .       , , ,     .     .  ,    .



: A Step by Step Backpropagation Eample by Matt Mazur

https://mattmazur.com/2015/03/17/a-step-by-step-backpropagation-example/ (https://mattmazur.com/2015/03/17/a-step-by-step-backpropagation-example/)



     !  ! .       .



,      -  ,         ,   . -, ,     , ..  ,    .    ,   .   ,   ,      .   ,     , .. .     ,             .



     .    (, ),      ,          .  ,   ,    ,  ,      ,      ,           ..     ()    ,    .  ,        .     ,      .



       .  ,      ,          .        return *;     !    !   ,        -   ,          ,    ",           !",    .      ,     .



!  ,         (    ).              . ..  ""  ,    .   !  :

 : (^2)' =2, ..  ""       1.

 : d(^2)= (^3)/3, ..  f()      ,     (  )

      : https://ru.wikipedia.org/wiki/_#_ (https://ru.wikipedia.org/wiki/%D0%9F%D1%80%D0%BE%D0%B8%D0%B7%D0%B2%D0%BE%D0%B4%D0%BD%D0%B0%D1%8F_%D1%84%D1%83%D0%BD%D0%BA%D1%86%D0%B8%D0%B8#%D0%A2%D0%B0%D0%B1%D0%BB%D0%B8%D1%86%D0%B0_%D0%BF%D1%80%D0%BE%D0%B8%D0%B7%D0%B2%D0%BE%D0%B4%D0%BD%D1%8B%D1%85)



      "         ".  ,   ,  -,  5    .



  .   0..9  MNIST.      ,  ,     .    .




  .


   .

   ,     (https://www.litres.ru/pages/biblio_book/?art=53604467)  .

      Visa, MasterCard, Maestro,    ,   ,     ,  PayPal, WebMoney, ., QIWI ,       .


