Triplet Loss 损失函数应用实战-完整代码+数据

# Triplet Loss 损失函数 [Triplet Loss](https://arxiv.org/pdf/1503.03832.pdf)是深度学习中的一种损失函数，用于训练**差异性较小**的样本，如人脸等， Feed数据包括锚（Anchor）示例、正（Positive）示例、负（Negative）示例，通过优化锚示例与正示例的距离**小于**锚示例与负示例的距离，实现样本的相似性计算。 <img src="doc/tl_algorithm.png" width=auto height="200"> 数据集：[MNIST](http://yann.lecun.com/exdb/mnist/) 框架：[DL-Project-Template](https://github.com/SpikeKing/DL-Project-Template) 目标：通过Triplet Loss训练模型，实现手写图像的相似性计算。 [工程](https://github.com/SpikeKing/triplet-loss-mnist)：https://github.com/SpikeKing/triplet-loss-mnist --- ## 模型 Triplet Loss的核心是锚示例、正示例、负示例共享模型，通过模型，将锚示例与正示例聚类，远离负示例。 **Triplet Loss Model**的结构如下： <img src="doc/tl_model.png" width=auto height="150"> - 输入：三个输入，即锚示例、正示例、负示例，不同示例的**结构**相同； - 模型：一个共享模型，支持替换为**任意**网络结构； - 输出：一个输出，即三个模型输出的拼接。 **Shared Model**选择常用的卷积模型，输出为全连接的128维数据： <img src="doc/base_model.png" width=auto height="400"> Triplet Loss **损失函数**的计算公式如下： <img src="doc/tl_formular.png" width=auto height="80"> --- ## 训练 模型参数： - batch_size：32 - epochs：2 超参数： - 边界Margin的值设置为``1``。 训练命令： ```text python main_train.py -c configs/triplet_config.json ``` 训练日志： ```text Using TensorFlow backend. [INFO] 解析配置... [INFO] 加载数据... [INFO] X_train.shape: (60000, 28, 28, 1), y_train.shape: (60000, 10) [INFO] X_test.shape: (10000, 28, 28, 1), y_test.shape: (10000, 10) [INFO] 构造网络... [INFO] model - 锚shape: (?, 128) [INFO] model - 正shape: (?, 128) [INFO] model - 负shape: (?, 128) [INFO] model - triplet_loss shape: (?, 1) [INFO] 训练网络... [INFO] trainer - 类别数: 10 Train on 54200 samples, validate on 8910 samples 2018-04-24 10:59:06.130952: I tensorflow/core/platform/cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA Epoch 1/2 54200/54200 [==============================] - 84s 2ms/step - loss: 0.6083 - val_loss: 0.0515 Epoch 2/2 54200/54200 [==============================] - 83s 2ms/step - loss: 0.0869 - val_loss: 0.0314 ``` 算法收敛较好，Loss线性下降： <img src="doc/loss_timeline.png" width=auto height="200"> TF Graph： <img src="doc/graph.png" width=auto height="400"> --- ## 验证 **算法效率**（TPS）: 每秒48163次 (0.0207625 ms/t) 测试命令： ```text python main_test.py -c configs/triplet_config.json ``` 测试日志： ```text Using TensorFlow backend. [INFO] 解析配置... [INFO] 加载数据... [INFO] 预测数据... 展示数据: (10000, 784) 展示标签: (10000,) 日志目录: /Users/wang/workspace/triplet-loss-mnist/experiments/triplet_mnist/logs/default 2018-04-24 11:02:07.874682: I tensorflow/core/platform/cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA [INFO] model - triplet_loss shape: (?, 1) __________________________________________________________________________________________________ Layer (type) Output Shape Param # Connected to ================================================================================================== anc_input (InputLayer) (None, 28, 28, 1) 0 __________________________________________________________________________________________________ pos_input (InputLayer) (None, 28, 28, 1) 0 __________________________________________________________________________________________________ neg_input (InputLayer) (None, 28, 28, 1) 0 __________________________________________________________________________________________________ model_1 (Model) (None, 128) 112096 anc_input[0][0] pos_input[0][0] neg_input[0][0] __________________________________________________________________________________________________ concatenate_1 (Concatenate) (None, 384) 0 model_1[1][0] model_1[2][0] model_1[3][0] ================================================================================================== Total params: 112,096 Trainable params: 112,096 Non-trainable params: 0 __________________________________________________________________________________________________ TPS: 272553.829381 (0.003669 ms) 验证结果结构: (10000, 128) 展示数据: (10000, 128) 展示标签: (10000,) 日志目录: /Users/wang/workspace/triplet-loss-mnist/experiments/triplet_mnist/logs/test [INFO] 预测完成... ``` MNIST验证集的效果： ``` bash [INFO] trainer - clz 0 [INFO] trainer - distance - min: -15.4567, max: 1.98611, avg: -6.50481 [INFO] acc: 0.996632996633 [INFO] trainer - clz 1 [INFO] trainer - distance - min: -13.09, max: 3.43779, avg: -6.66867 [INFO] acc: 0.99214365881 [INFO] trainer - clz 2 [INFO] trainer - distance - min: -14.2524, max: 2.49437, avg: -5.60508 [INFO] acc: 0.991021324355 [INFO] trainer - clz 3 [INFO] trainer - distance - min: -16.6555, max: 1.21776, avg: -6.32161 [INFO] acc: 0.995510662177 [INFO] trainer - clz 4 [INFO] trainer - distance - min: -14.193, max: 1.65427, avg: -5.90896 [INFO] acc: 0.991021324355 [INFO] trainer - clz 5 [INFO] trainer - distance - min: -14.1007, max: 2.01843, avg: -6.36086 [INFO] acc: 0.994388327722 [INFO] trainer - clz 6 [INFO] trainer - distance - min: -16.8953, max: 2.84421, avg: -8.43978 [INFO] acc: 0.995510662177 [INFO] trainer - clz 7 [INFO] trainer - distance - min: -16.6177, max: 3.49675, avg: -5.99822 [INFO] acc: 0.989898989899 [INFO] trainer - clz 8 [INFO] trainer - distance - min: -14.937, max: 3.38141, avg: -5.4424 [INFO] acc: 0.979797979798 [INFO] trainer - clz 9 [INFO] trainer - distance - min: -16.9519, max: 2.39112, avg: -5.93581 [INFO] acc: 0.985409652076 ``` 测试的MNIST分布： <img src="doc/default.png" width=auto height="400"> 输出的Triplet Loss MNIST分布： <img src="doc/test.png" width=auto height="400"> 本例仅仅使用2个Epoch，也没有特殊设置超参，实际效果仍有提升空间。 --- By C. L. Wang