{
  "id": 131566,
  "title": "model usage",
  "url": "/competitions/deepfake-detection-challenge/discussion/131566",
  "author_name": "",
  "post_date": "2020-02-20T12:49:25.900078Z",
  "votes": null,
  "comment_count": 6,
  "views": 0,
  "content": "<p>can i use this type of model created using tensorflow i am having a good amount of accuracy </p>\n\n<p>def baseline_model():</p>\n\n<pre><code>model = Sequential()\n\nmodel.add(BatchNormalization(input_shape=(128,128,3)))\n\nmodel.add(Conv2D(32, (3, 3), padding='same', activation='relu'))\n\nmodel.add(MaxPooling2D(pool_size=(2, 2), strides=(2,2)))\n\nmodel.add(Dropout(0.25))\nmodel.add(BatchNormalization())\nmodel.add(Conv2D(32, (3, 3), padding='same', activation='relu'))\nmodel.add(MaxPooling2D(pool_size=(2, 2)))\nmodel.add(Dropout(0.25))\nmodel.add(Flatten())\nmodel.add(Dense(128, activation='relu'))\nmodel.add(Dropout(0.5))\nmodel.add(Dense(2, activation='softmax'))\nreturn model\n</code></pre>",
  "messages": [
    {
      "id": "751723",
      "postDate": "02/20/2020 12:49:25",
      "content": "<p>can i use this type of model created using tensorflow i am having a good amount of accuracy </p>\n\n<p>def baseline_model():</p>\n\n<pre><code>model = Sequential()\n\nmodel.add(BatchNormalization(input_shape=(128,128,3)))\n\nmodel.add(Conv2D(32, (3, 3), padding='same', activation='relu'))\n\nmodel.add(MaxPooling2D(pool_size=(2, 2), strides=(2,2)))\n\nmodel.add(Dropout(0.25))\nmodel.add(BatchNormalization())\nmodel.add(Conv2D(32, (3, 3), padding='same', activation='relu'))\nmodel.add(MaxPooling2D(pool_size=(2, 2)))\nmodel.add(Dropout(0.25))\nmodel.add(Flatten())\nmodel.add(Dense(128, activation='relu'))\nmodel.add(Dropout(0.5))\nmodel.add(Dense(2, activation='softmax'))\nreturn model\n</code></pre>",
      "rawMarkdown": "can i use this type of model created using tensorflow i am having a good amount of accuracy \n\n\ndef baseline_model():\n\n    model = Sequential()\n\n    model.add(BatchNormalization(input_shape=(128,128,3)))\n\n    model.add(Conv2D(32, (3, 3), padding='same', activation='relu'))\n\n    model.add(MaxPooling2D(pool_size=(2, 2), strides=(2,2)))\n\n    model.add(Dropout(0.25))\n    model.add(BatchNormalization())\n    model.add(Conv2D(32, (3, 3), padding='same', activation='relu'))\n    model.add(MaxPooling2D(pool_size=(2, 2)))\n    model.add(Dropout(0.25))\n    model.add(Flatten())\n    model.add(Dense(128, activation='relu'))\n    model.add(Dropout(0.5))\n    model.add(Dense(2, activation='softmax'))\n    return model",
      "votes": null
    },
    {
      "id": "751834",
      "postDate": "02/20/2020 14:59:21",
      "content": "<p>You certainly could use it, though I am not sure how effective it will be... one the positive side I don't think it'll have a big problem with overfitting!</p>",
      "rawMarkdown": "You certainly could use it, though I am not sure how effective it will be... one the positive side I don't think it'll have a big problem with overfitting!",
      "votes": null
    },
    {
      "id": "751861",
      "postDate": "02/20/2020 15:23:01",
      "content": "<p>You'd be detecting only a small combination of 5x5 pixel features (2x 3x3px).\nMaybe that's enough ;)\nWhat's your accuracy and log loss?</p>",
      "rawMarkdown": "You'd be detecting only a small combination of 5x5 pixel features (2x 3x3px).\nMaybe that's enough ;)\nWhat's your accuracy and log loss?",
      "votes": null
    },
    {
      "id": "752809",
      "postDate": "02/21/2020 12:32:53",
      "content": "<p>i am no expert in data science that is why again asking it is using keras sequential api and keras layers are they allowed because i have not seen anyone using them yet will this type of model using lstm and cnn be accepted the code displayed is only for example to ask question</p>",
      "rawMarkdown": "i am no expert in data science that is why again asking it is using keras sequential api and keras layers are they allowed because i have not seen anyone using them yet will this type of model using lstm and cnn be accepted the code displayed is only for example to ask question",
      "votes": null
    },
    {
      "id": "752810",
      "postDate": "02/21/2020 12:33:37",
      "content": "<p>i am no expert in data science that is why again asking it is using keras sequential api and keras layers are they allowed because i have not seen anyone using them yet will this type of model using lstm and cnn be accepted the code displayed is only for example to ask question</p>",
      "rawMarkdown": "i am no expert in data science that is why again asking it is using keras sequential api and keras layers are they allowed because i have not seen anyone using them yet will this type of model using lstm and cnn be accepted the code displayed is only for example to ask question",
      "votes": null
    },
    {
      "id": "752916",
      "postDate": "02/21/2020 14:42:28",
      "content": "<p>Sure <a href=\"/rishabh12j\">@rishabh12j</a> you can use any open source library or any kind of model you can think of.\nPyTorch is probably to be the most popular framework in Kaggle competitions right now, but Keras is good too.\nThere are lots of interesting ideas and tips in the public kernels and discussions. You can look for the hottest and most voted posts for example.</p>",
      "rawMarkdown": "Sure @rishabh12j you can use any open source library or any kind of model you can think of.\nPyTorch is probably to be the most popular framework in Kaggle competitions right now, but Keras is good too.\nThere are lots of interesting ideas and tips in the public kernels and discussions. You can look for the hottest and most voted posts for example.",
      "votes": null
    },
    {
      "id": "754886",
      "postDate": "02/24/2020 07:07:49",
      "content": "<p>thank you</p>",
      "rawMarkdown": "thank you",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 751834,
      "author_name": "jamesphoward",
      "author_url": "",
      "post_date": "02/20/2020 14:59:21",
      "content": "<p>You certainly could use it, though I am not sure how effective it will be... one the positive side I don't think it'll have a big problem with overfitting!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 751861,
      "author_name": "hmendonca",
      "author_url": "",
      "post_date": "02/20/2020 15:23:01",
      "content": "<p>You'd be detecting only a small combination of 5x5 pixel features (2x 3x3px).\nMaybe that's enough ;)\nWhat's your accuracy and log loss?</p>",
      "votes": null,
      "replies": [
        {
          "id": 752809,
          "author_name": "rishabh12j",
          "author_url": "",
          "post_date": "02/21/2020 12:32:53",
          "content": "<p>i am no expert in data science that is why again asking it is using keras sequential api and keras layers are they allowed because i have not seen anyone using them yet will this type of model using lstm and cnn be accepted the code displayed is only for example to ask question</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 752916,
          "author_name": "hmendonca",
          "author_url": "",
          "post_date": "02/21/2020 14:42:28",
          "content": "<p>Sure <a href=\"/rishabh12j\">@rishabh12j</a> you can use any open source library or any kind of model you can think of.\nPyTorch is probably to be the most popular framework in Kaggle competitions right now, but Keras is good too.\nThere are lots of interesting ideas and tips in the public kernels and discussions. You can look for the hottest and most voted posts for example.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 754886,
          "author_name": "rishabh12j",
          "author_url": "",
          "post_date": "02/24/2020 07:07:49",
          "content": "<p>thank you</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 752810,
      "author_name": "rishabh12j",
      "author_url": "",
      "post_date": "02/21/2020 12:33:37",
      "content": "<p>i am no expert in data science that is why again asking it is using keras sequential api and keras layers are they allowed because i have not seen anyone using them yet will this type of model using lstm and cnn be accepted the code displayed is only for example to ask question</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "751723": "can i use this type of model created using tensorflow i am having a good amount of accuracy \n\n\ndef baseline_model():\n\n    model = Sequential()\n\n    model.add(BatchNormalization(input_shape=(128,128,3)))\n\n    model.add(Conv2D(32, (3, 3), padding='same', activation='relu'))\n\n    model.add(MaxPooling2D(pool_size=(2, 2), strides=(2,2)))\n\n    model.add(Dropout(0.25))\n    model.add(BatchNormalization())\n    model.add(Conv2D(32, (3, 3), padding='same', activation='relu'))\n    model.add(MaxPooling2D(pool_size=(2, 2)))\n    model.add(Dropout(0.25))\n    model.add(Flatten())\n    model.add(Dense(128, activation='relu'))\n    model.add(Dropout(0.5))\n    model.add(Dense(2, activation='softmax'))\n    return model",
    "751834": "You certainly could use it, though I am not sure how effective it will be... one the positive side I don't think it'll have a big problem with overfitting!",
    "751861": "You'd be detecting only a small combination of 5x5 pixel features (2x 3x3px).\nMaybe that's enough ;)\nWhat's your accuracy and log loss?",
    "752809": "i am no expert in data science that is why again asking it is using keras sequential api and keras layers are they allowed because i have not seen anyone using them yet will this type of model using lstm and cnn be accepted the code displayed is only for example to ask question",
    "752810": "i am no expert in data science that is why again asking it is using keras sequential api and keras layers are they allowed because i have not seen anyone using them yet will this type of model using lstm and cnn be accepted the code displayed is only for example to ask question",
    "752916": "Sure @rishabh12j you can use any open source library or any kind of model you can think of.\nPyTorch is probably to be the most popular framework in Kaggle competitions right now, but Keras is good too.\nThere are lots of interesting ideas and tips in the public kernels and discussions. You can look for the hottest and most voted posts for example.",
    "754886": "thank you"
  },
  "source": "meta"
}