├── .gitignore ├── CONTRIBUTING.md ├── DAE.py ├── LICENSE ├── MLP.py ├── README.md ├── SCAN.py ├── VAE.py ├── assets └── SCAN-model.png ├── conf ├── __init__.py └── config.py ├── main.py └── utils ├── deepmind_lab.py └── resnet_utils.py /.gitignore: -------------------------------------------------------------------------------- 1 | *__pycache__ 2 | local 3 | -------------------------------------------------------------------------------- /CONTRIBUTING.md: -------------------------------------------------------------------------------- 1 | # Contributing 2 | contributing guide is coming. 3 | -------------------------------------------------------------------------------- /DAE.py: -------------------------------------------------------------------------------- 1 | import tensorflow as tf 2 | 3 | 4 | class Encoder(object): 5 | def __init__(self, cfg): 6 | self.cfg = cfg 7 | 8 | def __call__(self, input): 9 | if self.cfg.DAENoiseType == "gaussian": # additive gaussian noise 10 | noise = self.cfg.scale * tf.random_normal(shape=input.get_shape()) 11 | input = input + noise 12 | elif self.cfg.DAENoiseType == "mask": # masking noise 13 | input = tf.nn.dropout(input, self.cfg.keep_prob) 14 | h1 = tf.contrib.layers.conv2d(input, num_outputs=32, kernel_size=4, 15 | strde=2, activation_fn=tf.nn.elu) 16 | h2 = tf.contrib.layers.conv2d(h1, num_outputs=32, kernel_size=4, 17 | strde=2, activation_fn=tf.nn.elu) 18 | h3 = tf.contrib.layers.conv2d(h2, num_outputs=64, kernel_size=4, 19 | strde=2, activation_fn=tf.nn.elu) 20 | h4 = tf.contrib.layers.conv2d(h3, num_outputs=64, kernel_size=4, 21 | strde=2, activation_fn=tf.nn.elu) 22 | 23 | return(h4) 24 | 25 | 26 | class Decoder(object): 27 | def __init__(self, cfg): 28 | self.cfg = cfg 29 | 30 | def __call__(self, input): 31 | input = tf.contrib.layers.fully_connected(input) 32 | h1 = tf.contrib.layers.conv2d_transpose(input, num_outputs=64, 33 | kernel_size=4, stride=2, 34 | activation_fn=tf.nn.elu) 35 | h2 = tf.contrib.layers.conv2d_transpose(h1, num_outputs=64, stride=2, 36 | kernel_size=4, 37 | activation_fn=tf.nn.elu) 38 | h3 = tf.contrib.layers.conv2d_transpose(h2, num_outputs=32, stride=2, 39 | kernel_size=4, 40 | activation_fn=tf.nn.elu) 41 | h4 = tf.contrib.layers.conv2d_transpose(h3, num_outputs=32, stride=2, 42 | kernel_size=4, 43 | activation_fn=tf.nn.elu) 44 | 45 | return(h4) 46 | 47 | 48 | if __name__ == "__main__": 49 | pass 50 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | Apache License 2 | Version 2.0, January 2004 3 | http://www.apache.org/licenses/ 4 | 5 | TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 6 | 7 | 1. Definitions. 8 | 9 | "License" shall mean the terms and conditions for use, reproduction, 10 | and distribution as defined by Sections 1 through 9 of this document. 11 | 12 | "Licensor" shall mean the copyright owner or entity authorized by 13 | the copyright owner that is granting the License. 14 | 15 | "Legal Entity" shall mean the union of the acting entity and all 16 | other entities that control, are controlled by, or are under common 17 | control with that entity. For the purposes of this definition, 18 | "control" means (i) the power, direct or indirect, to cause the 19 | direction or management of such entity, whether by contract or 20 | otherwise, or (ii) ownership of fifty percent (50%) or more of the 21 | outstanding shares, or (iii) beneficial ownership of such entity. 22 | 23 | "You" (or "Your") shall mean an individual or Legal Entity 24 | exercising permissions granted by this License. 25 | 26 | "Source" form shall mean the preferred form for making modifications, 27 | including but not limited to software source code, documentation 28 | source, and configuration files. 29 | 30 | "Object" form shall mean any form resulting from mechanical 31 | transformation or translation of a Source form, including but 32 | not limited to compiled object code, generated documentation, 33 | and conversions to other media types. 34 | 35 | "Work" shall mean the work of authorship, whether in Source or 36 | Object form, made available under the License, as indicated by a 37 | copyright notice that is included in or attached to the work 38 | (an example is provided in the Appendix below). 39 | 40 | "Derivative Works" shall mean any work, whether in Source or Object 41 | form, that is based on (or derived from) the Work and for which the 42 | editorial revisions, annotations, elaborations, or other modifications 43 | represent, as a whole, an original work of authorship. For the purposes 44 | of this License, Derivative Works shall not include works that remain 45 | separable from, or merely link (or bind by name) to the interfaces of, 46 | the Work and Derivative Works thereof. 47 | 48 | "Contribution" shall mean any work of authorship, including 49 | the original version of the Work and any modifications or additions 50 | to that Work or Derivative Works thereof, that is intentionally 51 | submitted to Licensor for inclusion in the Work by the copyright owner 52 | or by an individual or Legal Entity authorized to submit on behalf of 53 | the copyright owner. For the purposes of this definition, "submitted" 54 | means any form of electronic, verbal, or written communication sent 55 | to the Licensor or its representatives, including but not limited to 56 | communication on electronic mailing lists, source code control systems, 57 | and issue tracking systems that are managed by, or on behalf of, the 58 | Licensor for the purpose of discussing and improving the Work, but 59 | excluding communication that is conspicuously marked or otherwise 60 | designated in writing by the copyright owner as "Not a Contribution." 61 | 62 | "Contributor" shall mean Licensor and any individual or Legal Entity 63 | on behalf of whom a Contribution has been received by Licensor and 64 | subsequently incorporated within the Work. 65 | 66 | 2. Grant of Copyright License. Subject to the terms and conditions of 67 | this License, each Contributor hereby grants to You a perpetual, 68 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 69 | copyright license to reproduce, prepare Derivative Works of, 70 | publicly display, publicly perform, sublicense, and distribute the 71 | Work and such Derivative Works in Source or Object form. 72 | 73 | 3. Grant of Patent License. Subject to the terms and conditions of 74 | this License, each Contributor hereby grants to You a perpetual, 75 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 76 | (except as stated in this section) patent license to make, have made, 77 | use, offer to sell, sell, import, and otherwise transfer the Work, 78 | where such license applies only to those patent claims licensable 79 | by such Contributor that are necessarily infringed by their 80 | Contribution(s) alone or by combination of their Contribution(s) 81 | with the Work to which such Contribution(s) was submitted. If You 82 | institute patent litigation against any entity (including a 83 | cross-claim or counterclaim in a lawsuit) alleging that the Work 84 | or a Contribution incorporated within the Work constitutes direct 85 | or contributory patent infringement, then any patent licenses 86 | granted to You under this License for that Work shall terminate 87 | as of the date such litigation is filed. 88 | 89 | 4. Redistribution. You may reproduce and distribute copies of the 90 | Work or Derivative Works thereof in any medium, with or without 91 | modifications, and in Source or Object form, provided that You 92 | meet the following conditions: 93 | 94 | (a) You must give any other recipients of the Work or 95 | Derivative Works a copy of this License; and 96 | 97 | (b) You must cause any modified files to carry prominent notices 98 | stating that You changed the files; and 99 | 100 | (c) You must retain, in the Source form of any Derivative Works 101 | that You distribute, all copyright, patent, trademark, and 102 | attribution notices from the Source form of the Work, 103 | excluding those notices that do not pertain to any part of 104 | the Derivative Works; and 105 | 106 | (d) If the Work includes a "NOTICE" text file as part of its 107 | distribution, then any Derivative Works that You distribute must 108 | include a readable copy of the attribution notices contained 109 | within such NOTICE file, excluding those notices that do not 110 | pertain to any part of the Derivative Works, in at least one 111 | of the following places: within a NOTICE text file distributed 112 | as part of the Derivative Works; within the Source form or 113 | documentation, if provided along with the Derivative Works; or, 114 | within a display generated by the Derivative Works, if and 115 | wherever such third-party notices normally appear. The contents 116 | of the NOTICE file are for informational purposes only and 117 | do not modify the License. You may add Your own attribution 118 | notices within Derivative Works that You distribute, alongside 119 | or as an addendum to the NOTICE text from the Work, provided 120 | that such additional attribution notices cannot be construed 121 | as modifying the License. 122 | 123 | You may add Your own copyright statement to Your modifications and 124 | may provide additional or different license terms and conditions 125 | for use, reproduction, or distribution of Your modifications, or 126 | for any such Derivative Works as a whole, provided Your use, 127 | reproduction, and distribution of the Work otherwise complies with 128 | the conditions stated in this License. 129 | 130 | 5. Submission of Contributions. Unless You explicitly state otherwise, 131 | any Contribution intentionally submitted for inclusion in the Work 132 | by You to the Licensor shall be under the terms and conditions of 133 | this License, without any additional terms or conditions. 134 | Notwithstanding the above, nothing herein shall supersede or modify 135 | the terms of any separate license agreement you may have executed 136 | with Licensor regarding such Contributions. 137 | 138 | 6. Trademarks. This License does not grant permission to use the trade 139 | names, trademarks, service marks, or product names of the Licensor, 140 | except as required for reasonable and customary use in describing the 141 | origin of the Work and reproducing the content of the NOTICE file. 142 | 143 | 7. Disclaimer of Warranty. Unless required by applicable law or 144 | agreed to in writing, Licensor provides the Work (and each 145 | Contributor provides its Contributions) on an "AS IS" BASIS, 146 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or 147 | implied, including, without limitation, any warranties or conditions 148 | of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A 149 | PARTICULAR PURPOSE. You are solely responsible for determining the 150 | appropriateness of using or redistributing the Work and assume any 151 | risks associated with Your exercise of permissions under this License. 152 | 153 | 8. Limitation of Liability. In no event and under no legal theory, 154 | whether in tort (including negligence), contract, or otherwise, 155 | unless required by applicable law (such as deliberate and grossly 156 | negligent acts) or agreed to in writing, shall any Contributor be 157 | liable to You for damages, including any direct, indirect, special, 158 | incidental, or consequential damages of any character arising as a 159 | result of this License or out of the use or inability to use the 160 | Work (including but not limited to damages for loss of goodwill, 161 | work stoppage, computer failure or malfunction, or any and all 162 | other commercial damages or losses), even if such Contributor 163 | has been advised of the possibility of such damages. 164 | 165 | 9. Accepting Warranty or Additional Liability. While redistributing 166 | the Work or Derivative Works thereof, You may choose to offer, 167 | and charge a fee for, acceptance of support, warranty, indemnity, 168 | or other liability obligations and/or rights consistent with this 169 | License. However, in accepting such obligations, You may act only 170 | on Your own behalf and on Your sole responsibility, not on behalf 171 | of any other Contributor, and only if You agree to indemnify, 172 | defend, and hold each Contributor harmless for any liability 173 | incurred by, or claims asserted against, such Contributor by reason 174 | of your accepting any such warranty or additional liability. 175 | 176 | END OF TERMS AND CONDITIONS 177 | 178 | APPENDIX: How to apply the Apache License to your work. 179 | 180 | To apply the Apache License to your work, attach the following 181 | boilerplate notice, with the fields enclosed by brackets "{}" 182 | replaced with your own identifying information. (Don't include 183 | the brackets!) The text should be enclosed in the appropriate 184 | comment syntax for the file format. We also recommend that a 185 | file or class name and description of purpose be included on the 186 | same "printed page" as the copyright notice for easier 187 | identification within third-party archives. 188 | 189 | Copyright {2017} {Naturomics Liao} 190 | 191 | Licensed under the Apache License, Version 2.0 (the "License"); 192 | you may not use this file except in compliance with the License. 193 | You may obtain a copy of the License at 194 | 195 | http://www.apache.org/licenses/LICENSE-2.0 196 | 197 | Unless required by applicable law or agreed to in writing, software 198 | distributed under the License is distributed on an "AS IS" BASIS, 199 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 200 | See the License for the specific language governing permissions and 201 | limitations under the License. 202 | -------------------------------------------------------------------------------- /MLP.py: -------------------------------------------------------------------------------- 1 | import tensorflow as tf 2 | 3 | 4 | class Decoder(object): 5 | def __init__(self, cfg=None): 6 | self.cfg = cfg 7 | 8 | def __call__(self, input): 9 | with tf.varible_scope("decoder"): 10 | h1 = tf.contrib.layers.conv2d_transpose(input, num_outputs=64, 11 | kernel_size=4, stride=2) 12 | h2 = tf.contrib.layers.conv2d_transpose(h1, num_outputs=64, 13 | kernel_size=4, stride=2) 14 | h3 = tf.contrib.layers.conv2d_transpose(h2, num_outputs=32, 15 | kernel_size=4, stride=2) 16 | h4 = tf.contrib.layers.conv2d_transpose(h3, num_outputs=32, 17 | kernel_size=4, stride=2) 18 | 19 | return(h4) 20 | 21 | 22 | class Encoder(object): 23 | def __init__(self, cfg=None): 24 | self.cfg = cfg 25 | 26 | def __call__(self, input): 27 | """ 28 | Args: 29 | input: image Tensors. 30 | 31 | Returns: 32 | hidden: hidden layer. 33 | """ 34 | with tf.variable_scope("encoder"): 35 | hidden = tf.contrib.layers.fully_connected(input, num_outputs=100) 36 | 37 | return(hidden) 38 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # SCAN in Tensorflow 2 | 3 | [![Gitter](https://img.shields.io/gitter/room/nwjs/nw.js.svg?style=plastic)](https://gitter.im/SCAN-Tensorflow/Lobby) 4 | [![Contributions welcome](https://img.shields.io/badge/contributions-welcome-brightgreen.svg?style=plastic)](CONTRIBUTING.md) 5 | [![License](https://img.shields.io/badge/license-Apache%202.0-blue.svg?style=plastic)](https://opensource.org/licenses/Apache-2.0) 6 | 7 | A Tensorflow implementation of DeepMind's Symbol-Concept Association Network ([SCAN: Learning Abstract Hierarchical Compositional Visual Concepts](https://arxiv.org/abs/1707.03389)). 8 | 9 | SCAN is a framework for learning concepts paired with visual primitived. By teaching logical operators to SCAN, it even can learn to imagine new concepts from logical operators. See [the paper](https://arxiv.org/abs/1707.03389) and [DeepMind blog](https://deepmind.com/blog/imagine-creating-new-visual-concepts-recombining-familiar-ones/) for more details. 10 | 11 |

12 | SCAN model architecture 13 |

14 | 15 | # Usage 16 | 17 | ## Prerequisites 18 | 19 | * Python 3.4 20 | * DeepMind Lab 21 | * [Tensorflow 1.2](https://github.com/tensorflow/tensorflow/tree/r1.2) 22 | 23 | This is my configuration, other versions might work, too. If don't, let me know. 24 | 25 | It is worth mentioning that [the origin DeepMind Lab repo](https://github.com/deepmind/lab) hasn't been updated for months, and it doesn't support python3. On the other hand, I'm used to work on my Gentoo Linux with Python3. In order to make it work, I have forked this repo to [my github](https://github.com/naturomics/lab) and make it Gentoo and python3 supportable. If you are also using an unsupported platform or environment, see [here](https://github.com/naturomics/lab) for details. 26 | 27 | ## Instructions for runing 28 | ```shell 29 | $ git clone https://github.com/naturomics/SCAN-tensorflow.git 30 | $ cd SCAN-tensorflow 31 | ``` 32 | 33 | ```shell 34 | $ python main.py 35 | ``` 36 | 37 | # Results 38 | 39 | # Author 40 | Naturomics Liao - [@naturomics](https://github.com/naturomics) 41 | -------------------------------------------------------------------------------- /SCAN.py: -------------------------------------------------------------------------------- 1 | import tensorflow as tf 2 | 3 | import VAE 4 | import MLP 5 | 6 | 7 | # TF 1.3 release the statistical distribution library tf.distributions, 8 | # support for versions of TF before 1.3 9 | try: 10 | distributions = tf.distributions 11 | kl_divergence = tf.distributions.kl_divergence 12 | except: 13 | distributions = tf.contrib.distributions 14 | kl_divergence = tf.contrib.distributions.kl_divergence 15 | 16 | 17 | class SCAN(object): 18 | def __init__(self, cfg): 19 | self.cfg = cfg 20 | self.img_encoder = VAE.Encoder() 21 | self.img_decoder = VAE.Decoder() 22 | self.sym_encoder = MLP.Encoder() 23 | self.sym_decoder = MLP.Decoder() 24 | 25 | def train(self): 26 | img, sym = self.read_data_sets() 27 | 28 | with tf.variable_scope("beta_VAE"): 29 | img_q_mu, img_q_sigma = self.img_encoder(img) 30 | img_z = distributions.Normal(img_q_mu, img_q_sigma) 31 | img_gen = self.img_decoder(img_z.sample(self.cfg.batch_size)) 32 | 33 | img_reconstruct_error = tf.reduce_mean(img_gen) 34 | 35 | img_z_prior = distributions.Normal() 36 | KL_divergence = kl_divergence(img_z, img_z_prior) 37 | KL_divergence = self.cfg.beta_vae * KL_divergence 38 | 39 | loss = img_reconstruct_error - KL_divergence 40 | 41 | # train beta VAE 42 | optimizer = tf.train.AdamOptimizer(self.cfg.learning_rate) 43 | train_op = optimizer.minimize(loss) 44 | 45 | for step in range(self.cfg.epoch): 46 | self.sess.run(train_op) 47 | 48 | with tf.variable_scope("SCAN"): 49 | sym_q_mu, sym_q_sigma = self.sym_encoder(sym) 50 | sym_z = distributions.Normal(sym_q_mu, sym_q_sigma) 51 | self.sym_decoder(sym_z.sample(self.cfg.batch_size)) 52 | 53 | sym_reconstruct_error = tf.reduce_mean() 54 | 55 | sym_z_prior = distributions.Normal() 56 | beta_KL_divergence = kl_divergence(sym_z, sym_z_prior) 57 | beta_KL_divergence = self.cfg.beta_scan * beta_KL_divergence 58 | 59 | lambda_KL_divergence = kl_divergence(img_z, sym_z) 60 | 61 | loss = sym_reconstruct_error - beta_KL_divergence 62 | loss -= self.cfg.lambda_scan * lambda_KL_divergence 63 | 64 | # train SCAN 65 | optimizer = tf.train.AdamOptimizer(self.cfg.learning_rate) 66 | train_op = optimizer.minimize(loss) 67 | 68 | for step in range(self.cfg.epoch): 69 | self.sess.run(train_op) 70 | 71 | def inference(self): 72 | pass 73 | 74 | def read_data_sets(self): 75 | """ 76 | Returns: 77 | data queues of image and symbol. 78 | """ 79 | img, sym = [], [] 80 | 81 | return(img, sym) 82 | -------------------------------------------------------------------------------- /VAE.py: -------------------------------------------------------------------------------- 1 | import tensorflow as tf 2 | 3 | 4 | class Decoder(object): 5 | def __init__(self, cfg): 6 | self.cfg = cfg 7 | 8 | def __call__(self, input): 9 | with tf.varible_scope("decoder"): 10 | h1 = tf.contrib.layers.conv2d_transpose(input, num_outputs=64, 11 | kernel_size=4, stride=2) 12 | h2 = tf.contrib.layers.conv2d_transpose(h1, num_outputs=64, 13 | kernel_size=4, stride=2) 14 | h3 = tf.contrib.layers.conv2d_transpose(h2, num_outputs=32, 15 | kernel_size=4, stride=2) 16 | h4 = tf.contrib.layers.conv2d_transpose(h3, num_outputs=32, 17 | kernel_size=4, stride=2) 18 | 19 | return(h4) 20 | 21 | 22 | class Encoder(object): 23 | def __init__(self, cfg): 24 | self.cfg = cfg 25 | 26 | def __call__(self, input): 27 | """ 28 | Args: 29 | input: image Tensors. 30 | 31 | Returns: 32 | out: output. 33 | """ 34 | with tf.variable_scope("encoder"): 35 | h1 = tf.contrib.layers.conv2d(input, num_outputs=32, 36 | kernel_size=4, stride=2) 37 | h2 = tf.contrib.layers.conv2d(h1, num_outputs=32, 38 | kernel_size=4, stride=2) 39 | h3 = tf.contrib.layers.conv2d(h2, num_outputs=64, 40 | kernel_size=4, stride=2) 41 | h4 = tf.contrib.layers.conv2d(h3, num_outputs=64, 42 | kernel_size=4, stride=2) 43 | h5 = tf.contrib.layers.fully_connected(h4, num_outputs=256) 44 | 45 | return(h5) 46 | -------------------------------------------------------------------------------- /assets/SCAN-model.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/naturomics/SCAN-tensorflow/86b2c388a88d4b46872277a961bd0d1cd8c54f86/assets/SCAN-model.png -------------------------------------------------------------------------------- /conf/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/naturomics/SCAN-tensorflow/86b2c388a88d4b46872277a961bd0d1cd8c54f86/conf/__init__.py -------------------------------------------------------------------------------- /conf/config.py: -------------------------------------------------------------------------------- 1 | from __future__ import absolute_import 2 | from __future__ import division 3 | from __future__ import print_function 4 | 5 | import tensorflow as tf 6 | 7 | 8 | flags = tf.app.flags 9 | 10 | 11 | ########################### 12 | # model structure # 13 | ########################### 14 | 15 | 16 | ########################### 17 | # hyper parameter # 18 | ########################### 19 | flags.DEFINE_integer('batch_size', 100, 'batch size') 20 | flags.DEFINE_float('learning_rate', 0.0001, 'learning rate') 21 | flags.DEFINE_integer('epoch', 1000, 'epoch for training') 22 | 23 | # for DAE model 24 | flags.DEFINE_string("DAENoiseType", 'mask', "the method of adding noise in \ 25 | DAE model, including additive gaussian(gaussian) noise \ 26 | and masking noise(mask), default to masking noise") 27 | if flags.FLAGS.DAENoiseType == "gaussian": 28 | flags.DEFINE_float("scale", 0.5, "scale of gaussian noise") 29 | elif flags.FLAGS.DAENoiseType == "mask": 30 | flags.DEFINE_float("keep_prob", 0.8, "the keeping probability of dropout") 31 | else: 32 | tf.logging.error("check for the DAE noise method in the config file, \ 33 | it should be one of gaussian and mask") 34 | exit(1) 35 | 36 | # for VAE model 37 | flags.DEFINE_float('beta_vae', 50, 'hyperparameter beta for VAE objective') 38 | 39 | # for SCAN model 40 | flags.DEFINE_float('beta_scan', 1, 'hyperparameter beta for SCAN objective') 41 | flags.DEFINE_float('lambda_scan', 10, 'hyperparameter lambda \ 42 | for SCAN objective') 43 | 44 | 45 | ################################### 46 | # running environment setting # 47 | ################################### 48 | flags.DEFINE_integer('seed', 1178, 'seed for random number generation') 49 | flags.DEFINE_boolean('is_train', True, 'train or inference') 50 | -------------------------------------------------------------------------------- /main.py: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env python 2 | 3 | import tensorflow as tf 4 | from tensorflow import logging 5 | 6 | import conf.config 7 | from SCAN import SCAN 8 | 9 | FLAGS = tf.app.flags.FLAGS 10 | logging.set_verbosity(tf.logging.INFO) 11 | 12 | 13 | def main(_): 14 | with tf.Session() as sess: 15 | cfg = FLAGS 16 | scan = SCAN(sess, cfg) 17 | if FLAGS.is_train: 18 | logging.info("start training...") 19 | scan.train() 20 | logging.info("train finished.") 21 | else: 22 | scan.inference() 23 | 24 | 25 | if __name__ == "__main__": 26 | tf.app.run() 27 | -------------------------------------------------------------------------------- /utils/deepmind_lab.py: -------------------------------------------------------------------------------- 1 | import deepmind_lab 2 | 3 | 4 | class DeepMindLabAgent(object): 5 | def __init__(self, level, obsevations, config={}): 6 | # Construct and start the environment. 7 | env = deepmind_lab.Lab(level, obsevations, config) 8 | env.reset() 9 | 10 | def step(self, reward, unused_frame): 11 | pass 12 | 13 | 14 | if __name__ == '__main__': 15 | pass 16 | -------------------------------------------------------------------------------- /utils/resnet_utils.py: -------------------------------------------------------------------------------- 1 | import tensorflow as tf 2 | 3 | 4 | def preBlock(x): 5 | out = tf.layers.conv3d(x, filters=24, kernel_size=3, 6 | padding="same", name="preBlk.conv3d1") 7 | out = tf.layers.batch_normalization(out, axis=4, name="preBlk.batchNorm1") 8 | out = tf.nn.relu(out, name="preBlk.relu1") 9 | out = tf.layers.conv3d(out, filters=24, kernel_size=3, 10 | padding="same", name="preBlk.conv3d2") 11 | out = tf.layers.batch_normalization(out, axis=4, name="preBlk.batchNorm2") 12 | out = tf.nn.relu(out, name="preBlk.relu2") 13 | 14 | return(out) 15 | 16 | 17 | def block(x, n_filters, scope, stride=1): 18 | shortcut = x 19 | with tf.variable_scope(scope): 20 | out = tf.layers.conv3d(x, n_filters, kernel_size=3, 21 | strides=(stride, stride, stride), 22 | padding="same", name="blk.conv3d1") 23 | out = tf.layers.batch_normalization(out, axis=4, name="blk.batchNorm1") 24 | out = tf.nn.relu(out, name="blk.relu1") 25 | out = tf.layers.conv3d(x, n_filters, kernel_size=3, 26 | padding="same", name="blk.conv3d2") 27 | out = tf.layers.batch_normalization(out, axis=4, name="blk.batchNorm2") 28 | 29 | if stride != 1 or n_filters != x.get_shape().as_list()[-1]: 30 | with tf.variable_scope("shortcut"): 31 | shortcut = tf.layers.conv3d(shortcut, n_filters, kernel_size=1, 32 | strides=(stride, stride, stride), 33 | name="shortcut.conv3d") 34 | shortcut = tf.layers.batch_normalization(shortcut, 35 | axis=4, 36 | name="shortcut.bn") 37 | out = out + shortcut 38 | out = tf.nn.relu(out, name="blk.relu2") 39 | 40 | return(out) 41 | 42 | 43 | def conv3d(inputs, filters=16): 44 | feature = tf.layers.conv3d(inputs, filters, padding="same") 45 | return(feature) 46 | 47 | 48 | def fc(inputs, num_output_units): 49 | output = tf.contrib.layers.legacy_fully_connected(inputs, num_output_units) 50 | return(output) 51 | 52 | 53 | def batchNorm3d(inputs): 54 | output = tf.layers.batch_normalization(inputs) 55 | return(output) 56 | 57 | 58 | def activation(inputs): 59 | return(tf.nn.relu(inputs)) 60 | 61 | 62 | def max_pool(): 63 | pass 64 | --------------------------------------------------------------------------------