{
  "block": 106148506,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "7117.340559 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2026-05-18T04:29:18",
  "trx_id": "b28fa741e843ff665ed8620db9855fad50ddc640",
  "trx_in_block": 1,
  "virtual_op": 0
}

{
  "block": 105996275,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "4405.130154 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2026-05-12T21:00:57",
  "trx_id": "693594ebbaa680e8b13a4d558446d17ccbdf78df",
  "trx_in_block": 0,
  "virtual_op": 0
}

{
  "block": 105516038,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "7129.856315 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2026-04-26T03:43:36",
  "trx_id": "95aafbe49daf2f77a7ad239e8138962e9683b1b6",
  "trx_in_block": 10,
  "virtual_op": 0
}

{
  "block": 102865436,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "4446.676973 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2026-01-23T19:10:24",
  "trx_id": "6279c3e82468033a48aac06858812adc3b230252",
  "trx_in_block": 2,
  "virtual_op": 0
}

{
  "block": 101393214,
  "op": [
    "comment",
    {
      "author": "phernandez41",
      "body": "Content visibility in large language model assistants depends on several elements. Relevance to user queries, clarity of expression, and context awareness play key roles. Additionally, optimization for natural language processing algorithms affects how content is ranked and presented. Using tools like an <a href=\"https://signum.ai/visibility-in-chatgpt-like-tools/\">ai brand visibility tool</a> helps monitor and improve how content performs across AI platforms. Consistency and alignment with user intent also boost visibility, while outdated or ambiguous information may reduce chances of appearing prominently in AI-generated responses.",
      "json_metadata": "{\"links\":[\"https://signum.ai/visibility-in-chatgpt-like-tools/\"],\"app\":\"steemit/0.2\"}",
      "parent_author": "niki196",
      "parent_permlink": "artificial-intelligence-simplified",
      "permlink": "t6p3hw",
      "title": ""
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2025-12-03T13:15:33",
  "trx_id": "eae9b4c2e4da71621219f2352baa29f13aaf50e4",
  "trx_in_block": 4,
  "virtual_op": 0
}

{
  "block": 91311685,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "4610.896170 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2024-12-17T14:22:12",
  "trx_id": "89cee50724a7a4c4e05dd53b1a2076cfa1e8b31d",
  "trx_in_block": 4,
  "virtual_op": 0
}

{
  "block": 79865848,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "4780.029702 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2023-11-14T06:03:39",
  "trx_id": "3c23f799ba2a6414cca6fd0a3b938dd62e35591a",
  "trx_in_block": 3,
  "virtual_op": 0
}

{
  "block": 78360308,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "7716.938488 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2023-09-22T08:15:09",
  "trx_id": "a83c134e3a6fbde1164de717de62f08ec2af1227",
  "trx_in_block": 2,
  "virtual_op": 0
}

{
  "block": 76761770,
  "op": [
    "comment",
    {
      "author": "billshiphr",
      "body": "I'm sure that many people still have various concerns about AI, although personally I don't see anything wrong with using it, except for the advantages. By the way, you can read more about this in article https://anyforsoft.com/blog/ai-in-media-and-entertainment/. AI opens up new opportunities for media and entertainment.",
      "json_metadata": "{\"links\":[\"https://anyforsoft.com/blog/ai-in-media-and-entertainment/\"],\"app\":\"steemit/0.2\"}",
      "parent_author": "niki196",
      "parent_permlink": "artificial-intelligence-simplified",
      "permlink": "ryicbd",
      "title": ""
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2023-07-28T13:04:27",
  "trx_id": "b32dc121756e5a88f56a2985c1ad49b74b55e30c",
  "trx_in_block": 0,
  "virtual_op": 0
}

{
  "block": 69118392,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "7938.989926 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2022-11-03T16:00:00",
  "trx_id": "d5a27bf2b7c09dbc399c769b658c9a7b084a15b1",
  "trx_in_block": 5,
  "virtual_op": 0
}

{
  "block": 60821849,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "8159.097527 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2022-01-17T21:22:24",
  "trx_id": "a92564d6d735e99902b4fc56fc74dd3bf7f1f4ba",
  "trx_in_block": 35,
  "virtual_op": 0
}

{
  "block": 54612272,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "8343.291815 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2021-06-14T04:37:57",
  "trx_id": "5f2ff686ad619ac3bd24fc87c8316f18a311c586",
  "trx_in_block": 3,
  "virtual_op": 0
}

{
  "block": 49359586,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "8530.713789 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2020-12-11T14:51:54",
  "trx_id": "ae717c3e365bd26469c1cf8b0b1201a1c0ed5eec",
  "trx_in_block": 0,
  "virtual_op": 0
}

{
  "block": 49211130,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "1912.543513 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2020-12-06T08:28:15",
  "trx_id": "66666bbe7961a9f16bcc7d34b59a28bab588bc5b",
  "trx_in_block": 6,
  "virtual_op": 0
}

{
  "block": 49194671,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "8536.921643 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2020-12-05T18:29:33",
  "trx_id": "93b3e1a282f1755d21d39064b063d91835c8d5f0",
  "trx_in_block": 9,
  "virtual_op": 0
}

{
  "block": 48266906,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "1920.017158 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2020-11-02T23:22:06",
  "trx_id": "b164c70f92f43121b6012ba96e5bbfbbaa05091b",
  "trx_in_block": 7,
  "virtual_op": 0
}

{
  "block": 43221427,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "8739.727002 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2020-05-09T09:29:21",
  "trx_id": "aca9c89b89db6f13049a1f91ca3c99e96f84f60d",
  "trx_in_block": 5,
  "virtual_op": 0
}

{
  "block": 43198188,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "1953.311140 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2020-05-08T13:39:21",
  "trx_id": "879dd8e675ef2c177eb0f97a8b067a9c3c69bacb",
  "trx_in_block": 26,
  "virtual_op": 0
}

{
  "block": 42702268,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "8749.991896 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2020-04-20T21:36:48",
  "trx_id": "a53c39d5e63326f01e7b673a74408fca57b37308",
  "trx_in_block": 25,
  "virtual_op": 0
}

{
  "block": 38567160,
  "op": [
    "comment",
    {
      "author": "steemitboard",
      "body": "Congratulations @niki196! You received a personal award!\n\n<table><tr><td>https://steemitimages.com/70x70/http://steemitboard.com/@niki196/birthday2.png</td><td>Happy Birthday! - You are on the Steem blockchain for 2 years!</td></tr></table>\n\n<sub>_You can view [your badges on your Steem Board](https://steemitboard.com/@niki196) and compare to others on the [Steem Ranking](https://steemitboard.com/ranking/index.php?name=niki196)_</sub>\n\n\n###### [Vote for @Steemitboard as a witness](https://v2.steemconnect.com/sign/account-witness-vote?witness=steemitboard&approve=1) to get one more award and increased upvotes!",
      "json_metadata": "{\"image\":[\"https://steemitboard.com/img/notify.png\"]}",
      "parent_author": "niki196",
      "parent_permlink": "tensorflow-vs-pytorch-3-weeks-summary",
      "permlink": "steemitboard-notify-niki196-20191128t101315000z",
      "title": ""
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2019-11-28T10:13:15",
  "trx_id": "65baa4aa97881ea8ecb93f8ddd560e41ff4d7e8d",
  "trx_in_block": 6,
  "virtual_op": 0
}

{
  "block": 32981174,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "8945.520979 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2019-05-17T08:02:39",
  "trx_id": "ab79ef0b8e17056b277080906d29f4d3a8d1ad50",
  "trx_in_block": 56,
  "virtual_op": 0
}

{
  "block": 28102521,
  "op": [
    "comment",
    {
      "author": "steemitboard",
      "body": "Congratulations @niki196! You received a personal award!\n\n<table><tr><td>https://steemitimages.com/70x70/http://steemitboard.com/@niki196/birthday1.png</td><td>1 Year on Steemit</td></tr></table>\n\n<sub>_[Click here to view your Board of Honor](https://steemitboard.com/@niki196)_</sub>\n\n\n> Support [SteemitBoard's project](https://steemit.com/@steemitboard)! **[Vote for its witness](https://v2.steemconnect.com/sign/account-witness-vote?witness=steemitboard&approve=1)** and **get one more award**!",
      "json_metadata": "{\"image\":[\"https://steemitboard.com/img/notify.png\"]}",
      "parent_author": "niki196",
      "parent_permlink": "tensorflow-vs-pytorch-3-weeks-summary",
      "permlink": "steemitboard-notify-niki196-20181128t182006000z",
      "title": ""
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-11-28T18:20:06",
  "trx_id": "7233acc25fd6d68ff71031de4cb975bf259ae6ef",
  "trx_in_block": 8,
  "virtual_op": 0
}

{
  "block": 22646462,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "9144.997043 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-05-22T06:49:18",
  "trx_id": "7b51736a9dae03d5cb95f53239cdbfc73923f8d8",
  "trx_in_block": 13,
  "virtual_op": 0
}

{
  "block": 21771277,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "29535.848701 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-04-21T20:49:27",
  "trx_id": "ba73c0be4007a42c07d26d766ffd5efeeb09f252",
  "trx_in_block": 45,
  "virtual_op": 0
}

{
  "block": 21344996,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "de-clutter-your-life",
      "voter": "acarya.iam",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-04-07T01:32:18",
  "trx_id": "8c6e23bae193260346336bbea00fa24ad02794c7",
  "trx_in_block": 61,
  "virtual_op": 0
}

{
  "block": 20898943,
  "op": [
    "comment",
    {
      "author": "upv0t3",
      "body": "Hola @niki196, upv0t3\r\nEste es un servicio <b>gratuito</b> para nuevos usuarios de steemit, para apoyarlos y motivarlos a seguir generando contenido de valor para la comunidad.\r\n<3 Este es un corazón, o un helado, tu eliges\r.\r\n\n<h1> : ) </h1>\r\nN0. R4ND0M: \r\n6791 5378 3324 1381\n5654 6542 9602 2378\n1748 7405 8347 9808\n1487 4919 7216 9395\n",
      "json_metadata": "{\"tags\":[\"spanish\",\"kame\"],\"users\":[\"niki196\"],\"app\":\"null/null\",\"format\":\"markdown\"}",
      "parent_author": "niki196",
      "parent_permlink": "tensorflow-vs-pytorch-3-weeks-summary",
      "permlink": "re-niki196-tensorflow-vs-pytorch-3-weeks-summary-1521725204015t28e63fc9-482d-4648-acdf-be620c2c939fuid",
      "title": ""
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-03-22T13:26:45",
  "trx_id": "051fe24ebf118db84731dd2e651d8c9fc38823f0",
  "trx_in_block": 27,
  "virtual_op": 0
}

{
  "block": 20898942,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "tensorflow-vs-pytorch-3-weeks-summary",
      "voter": "upv0t3",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-03-22T13:26:42",
  "trx_id": "497335315cc2b346ebf38041846c70c4028f2b0f",
  "trx_in_block": 50,
  "virtual_op": 0
}

{
  "block": 20506009,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "tensorflow-vs-pytorch-3-weeks-summary",
      "voter": "dtubix",
      "weight": 5000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-03-08T21:19:30",
  "trx_id": "46a633ae1d18f0ff945dbcefbcac18c422c76e9d",
  "trx_in_block": 84,
  "virtual_op": 0
}

{
  "block": 20025701,
  "op": [
    "comment",
    {
      "author": "niki196",
      "body": "<html>\n<p>TensorFlow is developed by Google Brain and actively used at Google both for research and production needs. PyTorch is a cousin of lua-based Torch framework which is actively used at Facebook. PyTorch is dynamic computation graphs while Tensorflow is the static computation. &nbsp;&nbsp;&nbsp;</p>\n<p>&nbsp;<strong>Installation:</strong> &nbsp;</p>\n<ol>\n  <li>The installation is very easy and straightforward. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;PyTorch can be installed via PIP.</li>\n  <li>There isn't a designated CPU and GPU version of PyTorch &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;like there is with TensorFlow.</li>\n  <li>PyTorch does not offer an official window distribution &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;yet. There are non-official ports to windows, but there is no support from &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;PyTorch.</li>\n</ol>\n<p><strong>Data Loading:</strong> &nbsp;</p>\n<ol>\n  <li>API’s for data loading is well designed in PyTorch. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Parallelism in data loading is as simple as passing a num_workers argument &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;to the data loader. TensorFlow API’s are verbose and harder to learn</li>\n</ol>\n<p><strong>Documentation: </strong>&nbsp;&nbsp;</p>\n<ol>\n  <li>Documentation is complete for the most part. I never &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;failed to find the definition of function or module.</li>\n  <li>Opposed to Tensorflow, where all functions have one &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;page documentation, PyTorch uses one page one module. This is bit little &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;difficult.</li>\n</ol>\n<p><strong>Community:</strong> &nbsp;</p>\n<ol>\n  <li>Obviously, community is not as large as Tensorflow. &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Though many people enjoying working with PyTorch.</li>\n  <li>Community is large enough, questions on the official &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;forums gets quick answers.</li>\n</ol>\n<p><strong>Tools and Helpers:</strong> &nbsp;</p>\n<ol>\n  <li>PyTorch offers a fair amount of tools, some very useful &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;is still missing such as TensorFlow’s TensorBoard.</li>\n  <li>We can draw graphs with Matplotlib or seaborn libraries &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;in PyTorch. This needs a bit more self-written code than Tensorflow.</li>\n</ol>\n<p><strong>Deployment:</strong> &nbsp;</p>\n<ol>\n  <li>For small scale server deployment both are easy to wrap &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;in Flask web server. For heavily used machine learning services TensorFlow &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;is the winner.&nbsp;</li>\n  <li>For mobile deployment, TensorFlow works.</li>\n</ol>\n<p><strong>Device Management:</strong> &nbsp;</p>\n<ol>\n  <li>TensorFlow assumes you want to run on GPU if one is &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;available. In PyTorch you have to move explicitly everything onto the &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;device even if CUDA is enabled.&nbsp;</li>\n</ol>\n<p><strong>Debugging:</strong> &nbsp;</p>\n<ol>\n  <li>Since computation graph in PyTorch is defined at &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;runtime we can use our favorite Python debugging tools such as PyCharm &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;debugger or our old trusty print statement</li>\n  <li>This is not the case with TensorFlow. You have an &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;option to use a special tool called <a href=\"https://www.tensorflow.org/programmers_guide/debugger\">tfdbg</a> which allows to evaluate &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;TensorFlow expressions at runtime and browse all tensors and operations in &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;session scope. Of course, we won’t be able to debug any python code with &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;it.</li>\n  <li>Debugging Pytorch code is just like debugging python &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;code</li>\n</ol>\n<p><strong>Usage:</strong> &nbsp;</p>\n<ol>\n  <li>PyTorch &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;offers a very Pythonic API. In my opinion, this leads to more, but much &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;cleaner code.</li>\n  <li>PyTorch &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;graphs have to be defined in a class which inherits from the PyTorch &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;nn.Module class</li>\n  <li>A &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;forward() function gets called when the Graph is run.</li>\n  <li>This &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;new approach needs some time to get used to, but I think it is very intuitive &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;if you worked with Python outside the Deep learning before.</li>\n  <li>Based &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;on some reviews online, PyTorch also shows better performance on a lot of &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;model compared to Tensorflow.</li>\n</ol>\n<p>PyTorch is an awesome alternative to TensorFlow. Since PyTorch is still in Beta, I except some more changes and improvement to the usability, docs and performance. &nbsp;&nbsp;&nbsp;PyTorch is very pythonic and feels comfortable to work with it. It is also bit faster than TensorFlow. &nbsp;&nbsp;</p>\n</html>",
      "json_metadata": "{\"tags\":[\"tensorflow\",\"pytorch\",\"machinelearning\",\"deeplearning\"],\"links\":[\"https://www.tensorflow.org/programmers_guide/debugger\"],\"app\":\"steemit/0.1\",\"format\":\"html\"}",
      "parent_author": "",
      "parent_permlink": "tensorflow",
      "permlink": "tensorflow-vs-pytorch-3-weeks-summary",
      "title": "Tensorflow Vs Pytorch: 3 weeks summary"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-02-20T04:43:06",
  "trx_id": "f4b24a6cec115931c1cf2464eca8cdb950908b9e",
  "trx_in_block": 13,
  "virtual_op": 0
}

{
  "block": 20025536,
  "op": [
    "comment",
    {
      "author": "cheetah",
      "body": "Hi! I am a robot. I just upvoted you! I found similar content that readers might be interested in:\nhttps://towardsdatascience.com/fast-ai-v2-lesson1-synopsis-tl-dr-4985bba9eea2",
      "json_metadata": "",
      "parent_author": "niki196",
      "parent_permlink": "basics-of-ml-and-dl",
      "permlink": "cheetah-re-niki196basics-of-ml-and-dl",
      "title": ""
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-02-20T04:34:51",
  "trx_id": "378addce42a60645c5ec808f2e7b4e826dc25ef6",
  "trx_in_block": 57,
  "virtual_op": 0
}

{
  "block": 20025535,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "basics-of-ml-and-dl",
      "voter": "cheetah",
      "weight": 8
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-02-20T04:34:48",
  "trx_id": "d68950bf8dfd102227e854a4f2265faa5002984c",
  "trx_in_block": 29,
  "virtual_op": 0
}

{
  "block": 20025529,
  "op": [
    "comment",
    {
      "author": "niki196",
      "body": "<html>\n<p>If you are already familiar with the basics of DL and Machine Learning (ML), you can skip directly this entire post. But if you are new to this field, then the first few paragraphs are meant for you. As a Data Science newbie, there are a handful of things that you must know. &nbsp;&nbsp;</p>\n<p>Most importantly, <strong>the parameters</strong>. Parameters are all there is to a model. People talk a lot about parameters or weights (both are the same). Let’s say we have data which is already segregated into two classes. And the job of a model is to classify each data point into its respective bin. <strong>“The parameters”</strong> are those values that decide which point will go into which class. &nbsp;&nbsp;</p>\n<p><strong>The loss function </strong>is the next important thing that you should know. Loss function acts like a quality check for parameters. If the parameters are values which classify each data point, then loss function gives us the information about how good the parameters are for the given data. &nbsp;&nbsp;</p>\n<p>So <strong>training a model</strong> is nothing but finding the right parameters which give the least loss. We randomly take one set of parameters and update these after every iteration to get the minimum loss. <strong>Gradient descent </strong>is used to decide which direction takes us to that minimum. &nbsp;&nbsp;</p>\n<p>To give an analogy, if you are in a car at the top of a mountain, and the aim is to drive down, gradient descent is the direction in which your car travels. Speaking of driving a car downhill, there is one thing that we all do while driving downhill or rather, we MUST do. What might that be? Take a wild guess… Got it? <strong>YES!</strong> We use brakes. &nbsp;&nbsp;&nbsp;</p>\n<p>No sane person drives the car wildly without taking friction’s help. (except for a brake failure though). While updating the parameters using gradient descent, we apply a special kind of brakes, popularly known as <strong>the Learning rate.</strong> &nbsp;&nbsp;</p>\n<p>Learning rate provides a smooth transition from random parameters to state of the art models. Setting the learning rate too high is like accelerating the car downhill. Let alone reaching the foothill, the acceleration will throw you off the road. High learning rate too, will not converge to minimum loss. Loss increases with high learning rate and the model will stop learning. &nbsp;&nbsp;</p>\n<p>If the learning rate too small, it is like using the brakes extensively. The model will reach the minimum for sure. But it takes a lot of time for that to happen. And we cannot train a model for an indefinite amount of time. So setting the right learning rate is always very important. &nbsp;&nbsp;</p>\n<p>Finally, <strong>the parameter initialization.</strong> We always take a set of random numbers for parameters and update using gradient descent. <strong>Thumb rule for initialization: Symmetry does not work.</strong> Our brains are all the same. The first idea we get for initialization is a very simple one. “Why can’t we use all zeros?” Nope. That won’t work. &nbsp;&nbsp;</p>\n<p>The best way to initialize is to take random numbers from a <strong>Gaussian distribution</strong>. That just means taking random numbers and multiplying them by sqrt{2/n}. n is the total number of parameters across all the layers. &nbsp;&nbsp;&nbsp;&nbsp;</p>\n</html>",
      "json_metadata": "{\"tags\":[\"machinelearning\",\"deeplearning\"],\"app\":\"steemit/0.1\",\"format\":\"html\"}",
      "parent_author": "",
      "parent_permlink": "machinelearning",
      "permlink": "basics-of-ml-and-dl",
      "title": "Basics of ML and DL"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-02-20T04:34:30",
  "trx_id": "a0a3823565600a7c0a333d237e19775cb8331392",
  "trx_in_block": 28,
  "virtual_op": 0
}

{
  "block": 19967522,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "algorithm-storage-and-computation",
      "voter": "luiscordon",
      "weight": 500
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-02-18T04:12:51",
  "trx_id": "8c0d6c1e96782b38ccfbb937a36ebc64e98472f5",
  "trx_in_block": 16,
  "virtual_op": 0
}

{
  "block": 19959785,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "algorithm-storage-and-computation",
      "voter": "acostaana",
      "weight": 900
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-02-17T21:46:00",
  "trx_id": "97d84b10d2ca21a3e4c66bb19bc975b51bd06a7f",
  "trx_in_block": 50,
  "virtual_op": 0
}

{
  "block": 19419744,
  "op": [
    "author_reward",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "sbd_payout": "0.030 SBD",
      "steem_payout": "0.000 STEEM",
      "vesting_payout": "12.277373 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-30T03:05:00",
  "trx_id": "0000000000000000000000000000000000000000",
  "trx_in_block": 4294967295,
  "virtual_op": 2
}

{
  "block": 19229921,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "tineef",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T12:48:57",
  "trx_id": "7f9e08f7b7d8f032c304ece397b53455fb35f44c",
  "trx_in_block": 16,
  "virtual_op": 0
}

{
  "block": 19220576,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "stormi",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T05:01:33",
  "trx_id": "4e63f78f22e55ab99df171e01f72499214bb7b36",
  "trx_in_block": 38,
  "virtual_op": 0
}

{
  "block": 19220571,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "vintage078",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T05:01:18",
  "trx_id": "589205ffa874463c18fe5f3f7368dfc74c710d66",
  "trx_in_block": 43,
  "virtual_op": 0
}

{
  "block": 19220566,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "adabi",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T05:01:03",
  "trx_id": "12881135355c547050c89b53546079be4e91b628",
  "trx_in_block": 43,
  "virtual_op": 0
}

{
  "block": 19220565,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "tanyakoval",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T05:01:00",
  "trx_id": "ec09a9ddfee441398ad34fcab13817267a7f9c70",
  "trx_in_block": 28,
  "virtual_op": 0
}

{
  "block": 19220560,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "petyxovah14",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T05:00:45",
  "trx_id": "d5257e4b835d55b08120891f17ba208e95a66436",
  "trx_in_block": 33,
  "virtual_op": 0
}

{
  "block": 19220552,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "polac",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T05:00:21",
  "trx_id": "b69ed159059bd1f7a46d72c8395dc6c615bd7d64",
  "trx_in_block": 46,
  "virtual_op": 0
}

{
  "block": 19220533,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "belochka",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:59:24",
  "trx_id": "06f51a6d9bfb46f8fa33653a0bef3f1712542ffa",
  "trx_in_block": 47,
  "virtual_op": 0
}

{
  "block": 19220532,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "lovscan",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:59:21",
  "trx_id": "e199957c93101641061ed194ef5062a1a4bec4e5",
  "trx_in_block": 36,
  "virtual_op": 0
}

{
  "block": 19220532,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "shiasbatn",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:59:21",
  "trx_id": "d4ca62b63122345067e082901c0d68cba06f304b",
  "trx_in_block": 21,
  "virtual_op": 0
}

{
  "block": 19220529,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "balizovv",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:59:12",
  "trx_id": "3796bfbf8dc43f47081dbfb89f66009fbd26f165",
  "trx_in_block": 34,
  "virtual_op": 0
}

{
  "block": 19219959,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "ingurinka",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:30:42",
  "trx_id": "2668e2124d83534a25b925a010924243d8a289e9",
  "trx_in_block": 25,
  "virtual_op": 0
}

{
  "block": 19219955,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "jewnova",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:30:30",
  "trx_id": "497d50efcb441b1bdbc1d62b879d009a4b9ecd51",
  "trx_in_block": 42,
  "virtual_op": 0
}

{
  "block": 19219955,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "chrisencko",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:30:30",
  "trx_id": "e6414b492a2a235eaf8b6e0796d64d452facd594",
  "trx_in_block": 33,
  "virtual_op": 0
}

{
  "block": 19219952,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "sredinowa",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:30:21",
  "trx_id": "b2f4db458ec84b3262ef62bddbd7dcd0e0b0b87b",
  "trx_in_block": 22,
  "virtual_op": 0
}

{
  "block": 19219945,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "danilovvasiliyk",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:30:00",
  "trx_id": "3506357e64d4e7739b07c636c09e89ebae93b15f",
  "trx_in_block": 76,
  "virtual_op": 0
}

{
  "block": 19219945,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "uliauvarova8",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:30:00",
  "trx_id": "a9549d36ba9dafb80fa56e23bad9f1ea8fa9d0b8",
  "trx_in_block": 53,
  "virtual_op": 0
}

{
  "block": 19219941,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "greb",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:29:48",
  "trx_id": "423c74bc81776329895d85e892e3a35146715ea7",
  "trx_in_block": 17,
  "virtual_op": 0
}

{
  "block": 19219931,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "baryschewa",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:29:18",
  "trx_id": "18f7044ca193e1666d44d4193e71c9735fbe061f",
  "trx_in_block": 35,
  "virtual_op": 0
}

{
  "block": 19219926,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "mrmayo",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T04:29:03",
  "trx_id": "e6010ae4f1404e88b60d9a4f2fad66275cee6e6d",
  "trx_in_block": 25,
  "virtual_op": 0
}

{
  "block": 19218455,
  "op": [
    "comment",
    {
      "author": "niki196",
      "body": "<html>\n<p>There are 4 things to keep in mind when choosing or designing an algorithm for matrix computations: &nbsp;</p>\n<ol>\n  <li>Memory Use</li>\n  <li>Speed</li>\n  <li>Accuracy</li>\n  <li>Scalability/Parallelization</li>\n</ol>\n<p>Often there will be trade-offs between these categories. &nbsp;</p>\n<p><strong>Motivation: </strong>Matrices are everywhere-- anything that can be put in an <strong>Excel </strong>spreadsheet is a matrix, and <strong>language</strong> and <strong>pictures</strong> can be represented as matrices as well. Knowing what options there are for matrix algorithms, and how to navigate compromises, can make enormous differences to our solutions. For instance, an approximate matrix computation can often be thousands of times faster than an exact one. Knowing how the algorithms really work helps to both debug and accelerate our solution. &nbsp;&nbsp;<strong>Matrix Computations</strong> There are two key types of matrix computation, which get combined in many different ways. &nbsp;</p>\n<ol>\n  <li>Matrix &nbsp;and tensor products</li>\n  <li>Matrix &nbsp;decomposition's</li>\n</ol>\n<p>So basically we're going to be combining matrices, and pulling them apart again!</p>\n<p>&nbsp;<strong>“Math is continuous &amp; infinite, but computers are discrete &amp; finite.”</strong> &nbsp;</p>\n<p>&nbsp;Two Limitations of computer representations of numbers: &nbsp;</p>\n<ol>\n  <li>they &nbsp;can't be arbitrarily large or small</li>\n  <li>there &nbsp;must be gaps between them</li>\n</ol>\n<p>The reason we need to care about accuracy, is because computers can't store infinitely accurate numbers. It's possible to create calculations that give very wrong answers (particularly when repeating an operation many times, since each operation could multiply the error).&nbsp;</p>\n<p>&nbsp;<strong>Now, How computers store numbers:</strong> &nbsp;&nbsp;</p>\n<p><strong>IEEE Double precision arithmetic:</strong> Numbers can be as <strong>large</strong> as &nbsp;<strong>1.79 × 103081.79 × 10308</strong> &nbsp;and as <strong>small</strong> as &nbsp;<strong>2.2310−3082.23×10−308</strong>.&nbsp;</p>\n<p>The interval &nbsp;<strong>[1,2][1,2]</strong> &nbsp;is represented by discrete subset: 1, 1+2−52, 1+2×2−52, 1+3×2−52, …, 2 1, 1+2−52, 1+2×2−52, 1+3×2−52, …, 2 &nbsp;</p>\n<p><strong>Machine Epsilon</strong> Half the distance between 1 and the next larger number. This can vary by computer. <strong>IEEE </strong>standards for double precision specify&nbsp;</p>\n<p>Εmachine = 2−53 ≈ 1.11×10−16 εmachine= 2−53 ≈ 1.11×10−16 &nbsp;</p>\n<p>&nbsp;<strong>Two important properties of Floating Point Arithmetic:</strong> The difference between a real number &nbsp;x &nbsp;and its closest floating point approximation fl(x)fl(x) &nbsp;is always smaller than &nbsp;εmachineεmachine &nbsp;in relative terms. For some &nbsp;εε , where &nbsp;∣ε∣ ≤ εmachine ∣ε∣ ≤ εmachine , fl(x) = x⋅(1+ε) fl(x) = x⋅(1+ε)&nbsp;</p>\n<p>Where * is any operation ( +,−,×,÷+,−,×,÷ ), and &nbsp;⊛⊛ &nbsp;is its floating point analogue, x⊛y = (x∗y)(1+ε) x⊛y = (x∗y)(1+ε) &nbsp;&nbsp;for some &nbsp;εε , where &nbsp;∣ε∣≤εmachine∣ε∣≤εmachine &nbsp;That is, every operation of floating point arithmetic is exact up to a relative error of size at most &nbsp;εmachineεmachine.&nbsp;</p>\n<p>Since we cannot represent numbers exactly on a computer (due to the finiteness of our storage, and the gaps between numbers in floating point architecture), it becomes important to know how small perturbations in the input to a problem impact the output. &nbsp;</p>\n<p>&nbsp;<strong>\"A stable algorithm gives nearly the right answer to nearly the right question.\" --Trefethen</strong> &nbsp;&nbsp;</p>\n<p><strong>Conditioning:</strong> perturbation behavior of a mathematical problem (e.g. least squares) &nbsp;</p>\n<p><strong>Stability:</strong> perturbation behavior of an algorithm used to solve that problem on a computer (e.g. least squares algorithms, householder, back substitution, gaussian elimination) &nbsp;&nbsp;</p>\n<p><strong>Expensive Errors:</strong> The below examples are from <strong>Greenbaum &amp; Chartier</strong>. European Space Agency spent <strong>10 years and $7 billion</strong> on the Ariane 5 Rocket. What can happen when you try to fit a 64 bit number into a 16 bit space (integer overflow)</p>\n<p><strong>Sparse vs Dense:</strong> Now we know, how numbers are stored, now let's talk about how matrices are stored. A key way to save memory (and computation) is not to store all of your matrix. Instead, just store the non-zero elements. This is called <strong>sparse storage</strong>, and it is well suited to sparse matrices, that is, matrices where most elements are zero. &nbsp;&nbsp;&nbsp;</p>\n<p><strong>Speed</strong> differences come from a number of areas, particularly: &nbsp;</p>\n<ol>\n  <li>Computational &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;complexity</li>\n  <li>Vectorization</li>\n  <li>Scaling &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;to multiple cores and nodes</li>\n  <li>Locality</li>\n  <li>Computational &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;complexity</li>\n</ol>\n<p>Algorithms are generally expressed in terms of computation complexity with respect to the number of rows and number of columns in the matrix. E.g. you may find an algorithm described as (n2m)O(n2m) {Big O Notation} &nbsp;&nbsp;</p>\n<p><strong>Vectorization: </strong>Modern CPUs and GPUs can apply an operation to multiple elements at once on a single core. For instance, take the exponent of 4 floats in a vector in a single step. This is called <strong>SIMD</strong>. we will not be explicitly writing SIMD code (which tends to require assembly language or special C \"intrinsics\"), but instead will use vectorized operations in libraries like numpy, which in turn rely on specially tuned vectorized low level linear algebra APIs (in particular, BLAS, and LAPACK). &nbsp;&nbsp;</p>\n<p><strong>Locality</strong> Using slower ways to access data (e.g. over the internet) can be up to a billion times slower than faster ways (e.g. from a register). But there's much less fast storage than slow storage. So once we have data in fast storage, we want to do any computation required at that time, rather than having to load it multiple times each time we need it. In addition, for most types of storage its much faster to access data items that are stored next to each other, so we should try to always use any data stored nearby. These two issues are known as<strong> locality</strong>. &nbsp;&nbsp;</p>\n<p><strong>Speed of different types of memory</strong> Here are some numbers everyone should know (from the legendary <strong>Jeff Dean</strong>): &nbsp;</p>\n<p>&nbsp;L1 cache reference 0.5 ns&nbsp;</p>\n<p>L2 cache reference 7 ns&nbsp;</p>\n<p>Main memory reference/RAM 100 ns&nbsp;</p>\n<p>Send 2K bytes over 1 Gbps network 20,000 ns&nbsp;</p>\n<p>Read 1 MB sequentially from memory 250,000 ns</p>\n<p>&nbsp;Round trip within same datacenter 500,000 ns</p>\n<p>&nbsp;Disk seek 10,000,000 ns&nbsp;</p>\n<p>Read 1 MB sequentially from network 10,000,000 ns&nbsp;</p>\n<p>Read 1 MB sequentially from disk 30,000,000 ns</p>\n<p>Send packet CA-&gt;Netherlands-&gt;CA 150,000,000 ns &nbsp;</p>\n<p><strong>Key take-away:</strong> Each successive memory type is (at least) an order of magnitude worse than the one before it. Disk seeks are very slow. &nbsp;&nbsp;&nbsp;</p>\n</html>",
      "json_metadata": "{\"tags\":[\"datascience\",\"machinelearning\",\"algorithms\"],\"app\":\"steemit/0.1\",\"format\":\"html\"}",
      "parent_author": "",
      "parent_permlink": "datascience",
      "permlink": "algorithm-storage-and-computation",
      "title": "Algorithm storage and computation"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T03:15:27",
  "trx_id": "866ea1230122aa876ef6eeebc69c740fea8ad8fe",
  "trx_in_block": 36,
  "virtual_op": 0
}

{
  "block": 19218251,
  "op": [
    "comment",
    {
      "author": "cheetah",
      "body": "Hi! I am a robot. I just upvoted you! I found similar content that readers might be interested in:\nhttps://www.analyticsvidhya.com/blog/2017/06/word-embeddings-count-word2veec/",
      "json_metadata": "",
      "parent_author": "niki196",
      "parent_permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "permlink": "cheetah-re-niki1965i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "title": ""
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T03:05:15",
  "trx_id": "64593c4a6951292f5d5bcd9af1437c5725ea6b87",
  "trx_in_block": 32,
  "virtual_op": 0
}

{
  "block": 19218250,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "voter": "cheetah",
      "weight": 8
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T03:05:12",
  "trx_id": "ec8eb7c585d134db657c407d884285b83b7cd60c",
  "trx_in_block": 15,
  "virtual_op": 0
}

{
  "block": 19218246,
  "op": [
    "comment",
    {
      "author": "niki196",
      "body": "You possibly guess it right – **TEXT processing**. How do we make computers to perform clustering, classification etc. on a text data since we know that they are generally inefficient at handling and processing strings or texts for any fruitful outputs? \n\nSure, a computer can match two strings and tell you whether they are same or not. But how do we make computers tell you about football or Ronaldo when you search for Messi? How do you make a computer understand that “Apple” in “Apple is a tasty fruit” is a fruit that can be eaten and not a company?\n\nThe answer to the above question lie in creating a representation for words that capture their meanings, semantic relationships and the different types of contexts they are used in.\n\nAll of these are implemented by using **Word Embeddings** or numerical representation of texts, so that computers may handle them. Word Embeddings are the texts converted into numbers and there may be different numerical representations of the same text. \n\nOk, But - **Why we need Word Embedding**?\nMachine learning and Deep learning are incapable of processing strings as raw input. They requires numbers as their inputs to do any job Classification or Regression.\n\nHmm.. **Are there any types of Embedding?**\n1.\tFrequency Based Embedding: They are generally three types of Vectors\ni.\tCount Vector \nii.\tTF-IDF Vector\niii.\tCo-occurrence Vector\n2.\tPrediction based Embedding: \ni.\tCBOW (Continuous bag of words)\nii.\tSkip Gram model\n\n**Count Vector:** Consider a Corpus C of D documents {d1,d2…..dD} and N unique tokens extracted out of the corpus C. The N tokens will form our dictionary and the size of the Count Vector matrix M will be given by \nDXN. Each row in the matrix M contains the frequency of tokens in document D(i)\nExample: \nD1: He is lazy guy. She is lazy too\nD2: Nikita is lazy person\nDictionary of unique tokens: [‘He’,’lazy’,’guy’,’She’,Nikita,’person’]\nHere, D=2, N=6\n\n \tHe\tlazy\tguy\tShe\tNikita\tPerson\nD1\t1\t2\t1\t1\t0\t0\nD2\t0\t1\t0\t0\t1\t1\n\nNow, a column can also be understood as word vector for the corresponding word in the matrix M.\nFor example, the word vector for ‘lazy’ in the above matrix is [2,1] and so on. Here, the rows correspond to the documents in the corpus and the columns correspond to the tokens in the dictionary. The second row in the above matrix may be read as – D2 contains ‘lazy’: once, ‘Nikita’: once and ‘person’ once.\n\n**TF-IDF Vector:** This is another method which is based on the frequency method but it is different to the count vectorization in the sense that it takes into account not just the occurrence of a word in a single document but in the entire corpus.\n\nCommon words like ‘is’, ‘the’, ‘a’ etc. tend to appear quite frequently in comparison to the words which are important to a document. For example, a document A on Lionel Messi is going to contain more occurrences of the word “Messi” in comparison to other documents. But common words like “the” etc. are also going to be present in higher frequency in almost every document.\n\nWe would want is to down weight the common words occurring in almost all documents and give more importance to words that appear in a subset of documents.\n\nTF-IDF works by penalizing these common words by assigning them lower weights while giving importance to words that appear in a  subset of documents.\n\n**Co-occurrence Vector:** Similar words tend to occur together and will have similar context.\nFor example – Apple is a fruit. Mango is a fruit.\nApple and mango tend to have a similar context i.e. fruit\n\n**Co-occurrence Means** – For a given corpus, the co-occurrence of a pair of words say w1 and w2 is the number of times they have appeared together in a Context Window.\n\n**Context Window** –  Context window is specified by a number and the direction. \n\nLet’s say there are V unique words in the corpus. So Vocabulary size = V. The columns of the Co-occurrence matrix form the context words. Co-occurrence matrix is decomposed using techniques like PCA, SVD etc. into factors and combination of these factors forms the word vector representation.\n\n**Advantages of Co-occurrence Matrix:**\n1.\tIt preserves the semantic relationship between words. i.e. man and woman tend to be closer than man and apple.\n2.\tIt uses SVD at its core, which produces more accurate word vector representations than existing methods.\n3.\tIt has to be computed once and can be used anytime once computed. In this sense, it is faster in comparison to others.\n\n**Disadvantages of Co-Occurrence Matrix:**\nIt requires huge memory to store the co-occurrence matrix. But, this problem can be circumvented by factorizing the matrix out of the system for example in Hadoop clusters etc. and can be saved.\n\nhttps://github.com/Niki1ta96/Data-Science/blob/master/Python/Word%20Embedding%20basics.ipynb",
      "json_metadata": "{\"tags\":[\"machinelearning\",\"datascience\",\"textprocessing\",\"wordembedding\",\"frequencybasedembedding\"],\"links\":[\"https://github.com/Niki1ta96/Data-Science/blob/master/Python/Word%20Embedding%20basics.ipynb\"],\"app\":\"steemit/0.1\",\"format\":\"markdown\"}",
      "parent_author": "",
      "parent_permlink": "machinelearning",
      "permlink": "5i1axb-how-to-make-a-computer-to-understand-human-language-hmm",
      "title": "How to make a computer to understand human language..hmm..???"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2018-01-23T03:05:00",
  "trx_id": "c018825fb0b900dba4f1b2f285c8ea9a6d9739b5",
  "trx_in_block": 3,
  "virtual_op": 0
}

{
  "block": 18249770,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "machine-learning-humor",
      "voter": "nizam8",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-20T11:21:54",
  "trx_id": "7046b4591cd3e9d7e6d6d4128dfc73db112c7838",
  "trx_in_block": 5,
  "virtual_op": 0
}

{
  "block": 18249695,
  "op": [
    "comment",
    {
      "author": "niki196",
      "body": "There are two types of people \n\nFirst outliers : They are an inspiration\n\nSecond sampling errors : They need to be regularized. \n\n![sampling-error.jpg](https://steemitimages.com/DQmYwqGLRYUredbJziuxfzfxe8xzu2adjg5NC96cPsKrDfb/sampling-error.jpg)",
      "json_metadata": "{\"tags\":[\"life\",\"machinelearning\",\"humor\",\"laugh\"],\"image\":[\"https://steemitimages.com/DQmYwqGLRYUredbJziuxfzfxe8xzu2adjg5NC96cPsKrDfb/sampling-error.jpg\"],\"app\":\"steemit/0.1\",\"format\":\"markdown\"}",
      "parent_author": "",
      "parent_permlink": "life",
      "permlink": "machine-learning-humor",
      "title": "Machine Learning Humor"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-20T11:18:09",
  "trx_id": "bf43db1a6775b92b2cb7ab5c53973de4d24db485",
  "trx_in_block": 23,
  "virtual_op": 0
}

{
  "block": 18222138,
  "op": [
    "transfer",
    {
      "amount": "0.001 STEEM",
      "from": "hottopic",
      "memo": "Hello niki196. I Followed you.If you follow me, I'll be happy.Thanks :)",
      "to": "niki196"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-19T12:20:03",
  "trx_id": "37adb72403bc06a54d803427a0bd3748cd8aabe7",
  "trx_in_block": 3,
  "virtual_op": 0
}

{
  "block": 18222136,
  "op": [
    "comment",
    {
      "author": "niki196",
      "body": "“Feature engineering is another topic which doesn’t seem to merit any review papers or books, or even chapters in books, but it is absolutely vital to ML success. Much of the success of machine learning is actually success in engineering features.” — Scott Locklin, in “Neglected machine learning ideas”\n\nWhen our goal is to get the best possible results from a model, we need to get the most from what we have. But how do you get the most out of our data for modeling? This is the problem that the process and practice of feature engineering solves.\n\nOkay, What is feature engineering..?\n\nWhen we prepare a table for modeling, not all columns are useful in their raw form. In fact some columns (or attributes) may be useless - one example is the an ID type of attribute, for model building. \n\nFeature engineering as a technique, has three sub categories of techniques: feature selection, dimension reduction and feature generation.\n\nFeature Selection:\n\nThis is the process of ranking the attributes by their value to predictive ability of a model. Algorithms such as decision trees automatically rank the attributes in the data set. The top few nodes in a decision tree are considered the most important features from a predictive stand point. As a part of a process, feature selection using entropy based methods like decision trees can be employed to filter out less valuable attributes before feeding the reduced dataset to another modeling algorithm. Regression type models usually employ methods such as forward selection or backward elimination to select the final set of attributes for a model.\n\nDimension Reduction:\n\nThis is sometimes called feature extraction. The most classic example of dimension reduction is principle component analysis or PCA. PCA allows us to combine existing attributes into a new data frame consisting of a much reduced number of attributes by utilizing the variance in the data. The attributes which \"explain\" the highest amount of variance in the data form the first few principal components and we can ignore the rest of the attributes if data dimensionality is a problem from a computational standpoint. PCA results in a data table whose attributes do not look anything like the attributes of the raw dataset. \n\nFeature Generation or Feature Construction:\n\nThis technique is the one which most people are actually referring to when they talk about feature engineering. Quite simply, this is the process of manually constructing new attributes from raw data. It involves intelligently (a.k.a. domain knowledge) combining or splitting existing raw attributes into new one which have a higher predictive power. For example a date stamp may be used to generate 2 new attributes such as AM and PM which may be useful in discriminating whether day or night has a higher propensity to influence the response variable. We may want to convert noisy numerical attributes into simpler nominal attributes, by calculating the mean value and determining if a given row is above or below that mean value. We may generate a new attribute such as number of claims a member has filed for in a given time period, by combining date attribute and a nominal attribute such as claim_filed (Y/N), for example. The possibilities are endless. Feature construction is essentially a data transformation process.\n\nHere is a longer article on feature engineering which provides some excellent links and further readings for those who interested.\n\nhttp://machinelearningmastery.com/discover-feature-engineering-how-to-engineer-features-and-how-to-get-good-at-it/",
      "json_metadata": "{\"tags\":[\"featureengineering\",\"featureselection\",\"dimensionreduction\"],\"links\":[\"http://machinelearningmastery.com/discover-feature-engineering-how-to-engineer-features-and-how-to-get-good-at-it/\"],\"app\":\"steemit/0.1\",\"format\":\"markdown\"}",
      "parent_author": "",
      "parent_permlink": "featureengineering",
      "permlink": "feature-engineering",
      "title": "Feature Engineering"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-19T12:19:57",
  "trx_id": "dd5e0fa3cfd7001b2c2a6c94389e9486a803700f",
  "trx_in_block": 9,
  "virtual_op": 0
}

{
  "block": 18032582,
  "op": [
    "delegate_vesting_shares",
    {
      "delegatee": "niki196",
      "delegator": "steem",
      "vesting_shares": "29739.680753 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-12T22:19:57",
  "trx_id": "871880414549d43d34f7037684c3102ab8d07915",
  "trx_in_block": 16,
  "virtual_op": 0
}

{
  "block": 17941313,
  "op": [
    "author_reward",
    {
      "author": "niki196",
      "permlink": "apparently-kid-noah-ritter-i-love-him",
      "sbd_payout": "0.058 SBD",
      "steem_payout": "0.000 STEEM",
      "vesting_payout": "71.801201 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-09T18:15:21",
  "trx_id": "0000000000000000000000000000000000000000",
  "trx_in_block": 4294967295,
  "virtual_op": 7
}

{
  "block": 17876395,
  "op": [
    "curation_reward",
    {
      "comment_author": "neurallearner",
      "comment_permlink": "ai-joke-activities",
      "curator": "niki196",
      "reward": "6.155090 VESTS"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-07T12:07:03",
  "trx_id": "0000000000000000000000000000000000000000",
  "trx_in_block": 4294967295,
  "virtual_op": 5
}

{
  "block": 17838821,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "inspiring-nature",
      "voter": "niki196",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-06T04:47:18",
  "trx_id": "f16bdb2d0622fad08c4615ac2d89580181307efd",
  "trx_in_block": 1,
  "virtual_op": 0
}

{
  "block": 17836856,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "inspiring-nature",
      "voter": "anomaly",
      "weight": 100
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-06T03:09:03",
  "trx_id": "69d0ef1479a0ee6ffb92c745c9c0b78219e36bdc",
  "trx_in_block": 17,
  "virtual_op": 0
}

{
  "block": 17836229,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "inspiring-nature",
      "voter": "rekha007",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-06T02:37:42",
  "trx_id": "11616deb3401746896fda9d98f1d07d76d004c01",
  "trx_in_block": 19,
  "virtual_op": 0
}

{
  "block": 17836221,
  "op": [
    "comment",
    {
      "author": "niki196",
      "body": "https://www.youtube.com/watch?v=Pdrpv6h_jo4",
      "json_metadata": "{\"tags\":[\"life\",\"nature\",\"inspire\"],\"image\":[\"https://img.youtube.com/vi/Pdrpv6h_jo4/0.jpg\"],\"links\":[\"https://www.youtube.com/watch?v=Pdrpv6h_jo4\"],\"app\":\"steemit/0.1\",\"format\":\"markdown\"}",
      "parent_author": "",
      "parent_permlink": "life",
      "permlink": "inspiring-nature",
      "title": "Inspiring nature :)"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-06T02:37:18",
  "trx_id": "77b9a7c9e1e12c095c873cd7770797155ffc3219",
  "trx_in_block": 5,
  "virtual_op": 0
}

{
  "block": 17823923,
  "op": [
    "vote",
    {
      "author": "shagor0168",
      "permlink": "sh5-latest-cycle",
      "voter": "niki196",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-05T16:22:09",
  "trx_id": "327dbf7144248fd4e05dee5bcbfc77fbad645325",
  "trx_in_block": 32,
  "virtual_op": 0
}

{
  "block": 17822177,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "data-science-deep-learning-glossary",
      "voter": "nidhisteem",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-05T14:54:51",
  "trx_id": "4a62ba95314b0ea2f5682eb6650623159f65795c",
  "trx_in_block": 14,
  "virtual_op": 0
}

{
  "block": 17786951,
  "op": [
    "vote",
    {
      "author": "kongdong",
      "permlink": "colorchallenge-monday-red-red-poppies",
      "voter": "niki196",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-04T09:30:54",
  "trx_id": "45f7d8213c242e1745d89945f045183741845bc3",
  "trx_in_block": 4,
  "virtual_op": 0
}

{
  "block": 17766789,
  "op": [
    "vote",
    {
      "author": "niki196",
      "permlink": "exasperating-kid",
      "voter": "mikimasu",
      "weight": 10000
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-03T16:42:42",
  "trx_id": "2500114eab32607290c18c522e3832f78660d764",
  "trx_in_block": 18,
  "virtual_op": 0
}

{
  "block": 17766641,
  "op": [
    "comment",
    {
      "author": "niki196",
      "body": "https://www.youtube.com/watch?v=_yjlhlENiVo",
      "json_metadata": "{\"tags\":[\"life\",\"funny\",\"exasperatingkid\"],\"image\":[\"https://img.youtube.com/vi/_yjlhlENiVo/0.jpg\"],\"links\":[\"https://www.youtube.com/watch?v=_yjlhlENiVo\"],\"app\":\"steemit/0.1\",\"format\":\"markdown\"}",
      "parent_author": "",
      "parent_permlink": "life",
      "permlink": "exasperating-kid",
      "title": "Exasperating kid!!"
    }
  ],
  "op_in_trx": 0,
  "timestamp": "2017-12-03T16:35:18",
  "trx_id": "6bf1032d1dea20fc0ae299d0c2fd9d8e78c4519b",
  "trx_in_block": 30,
  "virtual_op": 0
}