{"id":20983874,"url":"https://github.com/j-w-yun/fizzbuzz_neural_network","last_synced_at":"2026-04-26T08:35:41.720Z","repository":{"id":136055407,"uuid":"122178726","full_name":"j-w-yun/fizzbuzz_neural_network","owner":"j-w-yun","description":"Approximate the FizzBuzz function using a neural network model in Tensorflow.","archived":false,"fork":false,"pushed_at":"2018-02-21T14:32:10.000Z","size":13,"stargazers_count":4,"open_issues_count":0,"forks_count":0,"subscribers_count":3,"default_branch":"master","last_synced_at":"2025-06-14T17:44:02.538Z","etag":null,"topics":["fizz-buzz","fizzbuzz","machine-learning","neural-network","tensorflow"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/j-w-yun.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null}},"created_at":"2018-02-20T09:36:44.000Z","updated_at":"2023-10-17T13:40:42.000Z","dependencies_parsed_at":"2023-12-07T13:00:18.901Z","dependency_job_id":null,"html_url":"https://github.com/j-w-yun/fizzbuzz_neural_network","commit_stats":null,"previous_names":["j-w-yun/fizzbuzz_neural_network"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/j-w-yun/fizzbuzz_neural_network","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/j-w-yun%2Ffizzbuzz_neural_network","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/j-w-yun%2Ffizzbuzz_neural_network/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/j-w-yun%2Ffizzbuzz_neural_network/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/j-w-yun%2Ffizzbuzz_neural_network/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/j-w-yun","download_url":"https://codeload.github.com/j-w-yun/fizzbuzz_neural_network/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/j-w-yun%2Ffizzbuzz_neural_network/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":32290896,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-26T08:29:33.829Z","status":"ssl_error","status_checked_at":"2026-04-26T08:29:18.366Z","response_time":129,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["fizz-buzz","fizzbuzz","machine-learning","neural-network","tensorflow"],"created_at":"2024-11-19T05:50:41.567Z","updated_at":"2026-04-26T08:35:41.692Z","avatar_url":"https://github.com/j-w-yun.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# fizzbuzz_neural_network\n## Approximating FizzBuzz\nI am approximating the infamous FizzBuzz function:\n\n    def fizzbuzz(start, end):\n        a = list()\n        for i in range(start, end + 1):\n            a.append(fb(i))\n        return a\n    \n    def fb(i):\n        if i % 3 == 0 and i % 5 == 0:\n            return \"FizzBuzz\"\n        elif i % 3 == 0:\n            return \"Fizz\"\n        elif i % 5 == 0:\n            return \"Buzz\"\n        else:\n            return i\n\nFrom 1 to 100, the correct output should be:\n\n    ['1' '2' 'Fizz' '4' 'Buzz' 'Fizz' '7' '8' 'Fizz' 'Buzz' '11' 'Fizz' '13'\n     '14' 'FizzBuzz' '16' '17' 'Fizz' '19' 'Buzz' 'Fizz' '22' '23' 'Fizz'\n     'Buzz' '26' 'Fizz' '28' '29' 'FizzBuzz' '31' '32' 'Fizz' '34' 'Buzz'\n     'Fizz' '37' '38' 'Fizz' 'Buzz' '41' 'Fizz' '43' '44' 'FizzBuzz' '46' '47'\n     'Fizz' '49' 'Buzz' 'Fizz' '52' '53' 'Fizz' 'Buzz' '56' 'Fizz' '58' '59'\n     'FizzBuzz' '61' '62' 'Fizz' '64' 'Buzz' 'Fizz' '67' '68' 'Fizz' 'Buzz'\n     '71' 'Fizz' '73' '74' 'FizzBuzz' '76' '77' 'Fizz' '79' 'Buzz' 'Fizz' '82'\n     '83' 'Fizz' 'Buzz' '86' 'Fizz' '88' '89' 'FizzBuzz' '91' '92' 'Fizz' '94'\n     'Buzz' 'Fizz' '97' '98' 'Fizz' 'Buzz']\n\nMy neural network is classifying each number into one of four categories:\n\n    0. Fizz\n    1. Buzz\n    2. FizzBuzz\n    3. None of the above\n\n## Required tools\n\n    import tensorflow as tf\n    import numpy as np\n\n## Preparing Data\nI am encoding the X (input) values as 16-bit binary:\n\n    def binary_encode_16b_array(a):\n        encoded_a = list()\n        for elem in a:\n            encoded_a.append(binary_encode_16b(elem))\n        return np.array(encoded_a)\n    \n    def binary_encode_16b(val):\n        bin_arr = list()\n        bin_str = format(val, '016b')\n        for bit in bin_str:\n            bin_arr.append(bit)\n        return np.array(bin_arr)\n\nAnd encoding the Y (output) values as one-hot vectors:\n\n    def one_hot_encode_array(a):\n        encoded_a = list()\n        for elem in a:\n            encoded_a.append(one_hot_encode(elem))\n        return np.array(encoded_a)\n    \n    def one_hot_encode(val):\n        if val == 'Fizz':\n            return np.array([1, 0, 0, 0])\n        elif val == 'Buzz':\n            return np.array([0, 1, 0, 0])\n        elif val == 'FizzBuzz':\n            return np.array([0, 0, 1, 0])\n        else:\n            return np.array([0, 0, 0, 1])\n\nwhich will categorize the 16-bit binary input data as one of the 4 possible categories specified by the FizzBuzz rule.\n\nFor example, if `[0.03 -0.4 -0.4  0.4]` is returned, the program knows not to print any of \"Fizz\", \"Buzz\", or \"FizzBuzz\":\n\n    # decoding values of Y\n    def one_hot_decode_array(x, y):\n        decoded_a = list()\n        for index, elem in enumerate(y):\n            decoded_a.append(one_hot_decode(x[index], elem))\n        return np.array(decoded_a)\n    \n    \n    def one_hot_decode(x, val):\n        index = np.argmax(val)\n        if index == 0:\n            return 'Fizz'\n        elif index == 1:\n            return 'Buzz'\n        elif index == 2:\n            return 'FizzBuzz'\n        elif index == 3:\n            return x\n\n## Initializing Data\nThis is how I am dividing up the training and testing data:\n\n    # train with data that will not be tested\n    test_x_start = 1\n    test_x_end = 100\n    train_x_start = 101\n    train_x_end = 10000\n    \n    test_x_raw = np.arange(test_x_start, test_x_end + 1)\n    test_x = binary_encode_16b_array(test_x_raw).reshape([-1, 16])\n    test_y_raw = fizzbuzz(test_x_start, test_x_end)\n    test_y = one_hot_encode_array(test_y_raw)\n    \n    train_x_raw = np.arange(train_x_start, train_x_end + 1)\n    train_x = binary_encode_16b_array(train_x_raw).reshape([-1, 16])\n    train_y_raw = fizzbuzz(train_x_start, train_x_end)\n    train_y = one_hot_encode_array(train_y_raw)\n\nso the model trains using values between 101 and 10000 and tests using values between 1 and 100.\n\n## Neural Network Model\nMy model architecture is simple, with 100 hidden neurons in one layer:\n\n    # define params\n    input_dim = 16\n    output_dim = 4\n    h1_dim = 100\n\n    # build graph\n    X = tf.placeholder(tf.float32, [None, input_dim])\n    Y = tf.placeholder(tf.float32, [None, output_dim])\n    \n    h1_w = tf.Variable(tf.random_normal([input_dim, h1_dim], stddev=0.1))\n    h1_b = tf.Variable(tf.zeros([h1_dim]))\n    h1_z = tf.nn.relu(tf.matmul(X, h1_w) + h1_b)\n    \n    fc_w = tf.Variable(tf.random_normal([h1_dim, output_dim], stddev=0.1))\n    fc_b = tf.Variable(tf.zeros([output_dim]))\n    Z = tf.matmul(h1_z, fc_w) + fc_b\n    \n    # define cost\n    cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=Y, logits=Z))\n    \n    # define op\n    train_step = tf.train.AdamOptimizer(0.001).minimize(cross_entropy)\n    \n    # define accuracy\n    correct_prediction = tf.equal(tf.argmax(Z, 1), tf.argmax(Y, 1))\n    correct_prediction = tf.cast(correct_prediction, tf.float32)\n    accuracy = tf.reduce_mean(correct_prediction)\n\n## Running the Model\nFor the sake of simplicity, I opted to omit batch training:\n\n    with tf.Session() as sess:\n        sess.run(tf.global_variables_initializer())\n    \n        for i in range(10000):\n            sess.run(train_step, feed_dict={X: train_x, Y: train_y})\n    \n            train_accuracy = sess.run(accuracy, feed_dict={X: train_x, Y: train_y})\n            print(i, \":\", train_accuracy)\n    \n        output = sess.run(Z, feed_dict={X: test_x})\n        decoded = one_hot_decode_array(test_x_raw, output)\n        print(decoded)\n\n## Results\nAfter about 5,000 iterations of train step, training accuracy converges to 1.0. Here is the following test output of the neural network model after 10,000 iterations of training:\n\n        0 : 0.346061\n        1 : 0.459596\n        2 : 0.48404\n        3 : 0.472828\n        4 : 0.441515\n        5 : 0.417071\n        \n        ...\n        \n        9998 : 1.0\n        9999 : 1.0\n        ['1' '2' 'Fizz' '4' 'Buzz' 'Fizz' '7' '8' 'Fizz' 'Buzz' '11' 'Fizz' '13'\n         '14' 'FizzBuzz' '16' '17' 'Fizz' '19' 'Buzz' 'Fizz' '22' '23' 'Fizz'\n         'Buzz' '26' 'Fizz' '28' '29' 'FizzBuzz' '31' '32' 'Fizz' '34' 'Buzz'\n         'Fizz' '37' '38' 'Fizz' 'Buzz' '41' 'Fizz' '43' '44' 'FizzBuzz' '46' '47'\n         'Fizz' '49' 'Buzz' 'Fizz' '52' '53' 'Fizz' 'Buzz' '56' 'Fizz' '58' '59'\n         'FizzBuzz' '61' '62' 'Fizz' '64' 'Buzz' 'Fizz' '67' '68' 'Fizz' 'Buzz'\n         '71' 'Fizz' '73' '74' 'FizzBuzz' '76' '77' 'Fizz' '79' 'Buzz' 'Fizz' '82'\n         '83' 'Fizz' 'Buzz' '86' 'Fizz' '88' '89' 'FizzBuzz' '91' '92' 'Fizz' '94'\n         'Buzz' 'Fizz' '97' '98' 'Fizz' 'Buzz']\n\nThis feedforward neural network model, despite its simplicity, without a priori knowledge of the modulo operation, successfully extrapolated the output of the FizzBuzz function in the domain that was excluded in its training data.\n\n[Go to Part 2 : Improving Model Accuracy][1].\n\n\n  [1]: https://github.com/Jaewan-Yun/fizzbuzz_neural_network/tree/master/Part%202%20-%20Improving%20Model%20Accuracy\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fj-w-yun%2Ffizzbuzz_neural_network","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fj-w-yun%2Ffizzbuzz_neural_network","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fj-w-yun%2Ffizzbuzz_neural_network/lists"}