remove(example/cifar100) remove the annotation code and chinese annot…

…ation

Signed-off-by: Marchons <[email protected]>

examples/cifar100/fci_ssl/fed_ci_match/algorithm/FedCiMatch.py

@@ -22,12 +22,9 @@
 
 
 def get_one_hot(target, num_classes):
-    # print(f'in get  one hot, target shape is {target.shape}')
     y = tf.one_hot(target, depth=num_classes)
-    # print(f'in get  one hot, y shape is {y.shape}')
     if len(y.shape) == 3:
         y = tf.squeeze(y, axis=1)
-    # print(f'in get  one hot, after tf.squeeze y shape is {y.shape}')
     return y
 
 
@@ -72,16 +69,12 @@ def build_feature_extractor(self):
 
         feature_extractor.build(input_shape=(None, 32, 32, 3))
         feature_extractor.call(keras.Input(shape=(32, 32, 3)))
-        # feature_extractor.load_weights(
-        #     "examples/cifar100/fci_ssl/fed_ci_match/algorithm/feature_extractor.weights.h5"
-        # )
         return feature_extractor
 
     def build_classifier(self):
         if self.classifier != None:
             new_classifier = keras.Sequential(
                 [
-                    # tf.keras.Input(shape=(None, self.feature_extractor.layers[-2].output_shape[-1])),
                     keras.layers.Dense(
                         self.num_classes, kernel_initializer="lecun_normal"
                     )
@@ -106,7 +99,6 @@ def build_classifier(self):
             )
             self.classifier = keras.Sequential(
                 [
-                    # tf.keras.Input(shape=(None, self.feature_extractor.layers[-2].output_shape[-1])),
                     keras.layers.Dense(
                         self.num_classes, kernel_initializer="lecun_normal"
                     )
@@ -192,10 +184,8 @@ def get_train_loader(self):
         )
         if len(self.exemplar_set) != 0:
             for exm_set in self.exemplar_set:
-                # print('in get train loader' , exm_set[0].shape)
                 train_x = np.concatenate((train_x, exm_set[0]), axis=0)
                 label = np.array(exm_set[1])
-                # label = label.reshape(-1, 1)
                 train_y = np.concatenate((train_y, label), axis=0)
             logging.info(
                 f"train_x shape: {train_x.shape} and train_y shape: {train_y.shape}"
@@ -250,8 +240,6 @@ def construct_exemplar_set(self, images, class_id, m):
         exemplar_data = []
         exemplar_label = []
         class_mean, fe_ouput = self.compute_exemplar_mean(images)
-        # cl_mean = tf.reduce_mean(class_mean, axis=0)
-        # now_class_mean = np.zeros((1, fe_ouput.shape[1]))
         diff = tf.abs(fe_ouput - class_mean)
         distance = [float(tf.reduce_sum(dis).numpy()) for dis in diff]
 
@@ -261,14 +249,7 @@ def construct_exemplar_set(self, images, class_id, m):
         exemplar_data = [images[i] for i in sorted_index]
         exemplar_label = [class_id] * len(exemplar_data)
         self.exemplar_set.append((exemplar_data, exemplar_label))
-        # for i in range(m):
-        #     x = class_mean - (now_class_mean + fe_ouput) / (i + 1)
-        #     x = np.linalg.norm(x)
-        #     index = np.argmin(x)
-        #     now_class_mean += fe_ouput[index]
-        #     exemplar_data.append(images[index])
-        #     exemplar_label.append(class_id)
-        # self.exemplar_set.append((exemplar_data, exemplar_label))
+
 
     def compute_exemplar_mean(self, images):
         images_data = (
@@ -279,7 +260,6 @@ def compute_exemplar_mean(self, images):
         fe_output = self.feature_extractor.predict(images_data)
         print("fe_output shape:", fe_output.shape)
         class_mean = tf.reduce_mean(fe_output, axis=0)
-        # class_mean = np.mean(fe_output, axis=0)
         return class_mean, fe_output
 
     def train(self, round):
@@ -291,11 +271,8 @@ def train(self, round):
         )
         q = []
         logging.info(f"is new task: {self.is_new_task}")
-        # if self.classifier is not None:
-        #     q = self.caculate_pre_update()
         if self.is_new_task:
             self.build_classifier()
-            # self.is_new_task = False
         all_params = []
         all_params.extend(self.feature_extractor.trainable_variables)
         all_params.extend(self.classifier.trainable_variables)

examples/cifar100/fci_ssl/fed_ci_match/algorithm/basemodel.py

@@ -38,7 +38,6 @@ def __init__(self, **kwargs) -> None:
         self.batch_size = kwargs.get("batch_size", 32)
         self.task_size = kwargs.get("task_size", 2)
         self.memory_size = kwargs.get("memory_size", 2000)
-        # self.fe = self.build_feature_extractor()
         self.num_classes = 50  # the number of class for the first task
         self.FedCiMatch = FedCiMatch(
             self.num_classes,

examples/cifar100/fci_ssl/fed_ci_match_v2/algorithm/FedCiMatch.py

@@ -75,7 +75,6 @@ def _build_classifier(self):
         if self.classifier != None:
             new_classifier = keras.Sequential(
                 [
-                    # tf.keras.Input(shape=(None, self.feature_extractor.layers[-2].output_shape[-1])),
                     keras.layers.Dense(
                         len(self.class_mapping), kernel_initializer="lecun_normal"
                     )
@@ -100,7 +99,6 @@ def _build_classifier(self):
             )
             self.classifier = keras.Sequential(
                 [
-                    # tf.keras.Input(shape=(None, self.feature_extractor.layers[-2].output_shape[-1])),
                     keras.layers.Dense(
                         len(self.class_mapping), kernel_initializer="lecun_normal"
                     )
@@ -126,7 +124,6 @@ def get_data_size(self):
     def model_call(self, x, training=False):
         x = self.feature_extractor(x, training=training)
         x = self.classifier(x, training=training)
-        # x = tf.nn.softmax(x)
         return x
 
     def _build_class_mapping(self):
@@ -181,7 +178,6 @@ def train(self, task_id, round):
 
         for epoch in range(self.epochs):
             for x, y in self.labeled_train_loader:
-                # logging.info(f' y is {y.numpy()}')
                 y = np.array([self.class_mapping[i] for i in y.numpy()])
                 tasks = self._split_tasks(x, y)
                 base_model_weights = self.feature_extractor.get_weights()

examples/cifar100/fci_ssl/fedavg/algorithm/basemodel.py

@@ -86,7 +86,6 @@ def build_classifier(self):
         if self.classifier != None:
             new_classifier = keras.Sequential(
                 [
-                    # tf.keras.Input(shape=(None, self.feature_extractor.layers[-2].output_shape[-1])),
                     keras.layers.Dense(
                         self.num_classes, kernel_initializer="lecun_normal"
                     )
@@ -109,7 +108,6 @@ def build_classifier(self):
             logging.info(f"input shape is {self.fe.layers[-2].output_shape[-1]}")
             self.classifier = keras.Sequential(
                 [
-                    # tf.keras.Input(shape=(None, self.feature_extractor.layers[-2].output_shape[-1])),
                     keras.layers.Dense(
                         self.num_classes, kernel_initializer="lecun_normal"
                     )
@@ -174,7 +172,7 @@ def predict(self, data_files, **kwargs):
             pred = tf.argmax(prob, axis=1)
             pred = tf.cast(pred, dtype=tf.int32)
             result[data] = pred.numpy()
-        print("finish predict")
+        logging.info("finish predict")
         return result
 
     def eval(self, data, round, **kwargs):

examples/cifar100/fci_ssl/fedavg/algorithm/resnet.py

@@ -55,7 +55,6 @@ def call(self, inputs, training=None):
         return output
 
 
-# 残差神经网络
 class ResNet(keras.Model):
     def __init__(self, layer_dims, num_classes=10):  # [2, 2, 2, 2]
         super(ResNet, self).__init__()
@@ -92,8 +91,6 @@ def call(self, inputs, training=None):
 
         # [b, c]
         x = self.avgpool(x)
-        # [b, 100]
-        # x = self.fc(x)
         return x
 
     def build_resblock(self, filter_num, blocks, stride=1):

examples/cifar100/fci_ssl/glfc/algorithm/GLFC.py

@@ -23,12 +23,9 @@
 
 
 def get_one_hot(target, num_classes):
-    # print(f'in get  one hot, target shape is {target.shape}')
     y = tf.one_hot(target, depth=num_classes)
-    # print(f'in get  one hot, y shape is {y.shape}')
     if len(y.shape) == 3:
         y = tf.squeeze(y, axis=1)
-    # print(f'in get  one hot, after tf.squeeze y shape is {y.shape}')
     return y
 
 
@@ -87,7 +84,6 @@ def _initialize_classifier(self):
         if self.classifier != None:
             new_classifier = tf.keras.Sequential(
                 [
-                    # tf.keras.Input(shape=(None, self.feature_extractor.layers[-2].output_shape[-1])),
                     tf.keras.layers.Dense(
                         self.num_classes, kernel_initializer="lecun_normal"
                     )
@@ -112,7 +108,6 @@ def _initialize_classifier(self):
             )
             self.classifier = tf.keras.Sequential(
                 [
-                    # tf.keras.Input(shape=(None, self.feature_extractor.layers[-2].output_shape[-1])),
                     tf.keras.layers.Dense(
                         self.num_classes, kernel_initializer="lecun_normal"
                     )
@@ -163,7 +158,6 @@ def update_new_set(self, need_update):
             self._reduce_exemplar_set(m)
             for i in self.last_class:
                 images = self.get_train_set_data(i)
-                # print(f'process class {i} with {len(images)} images')
                 self._construct_exemplar_set(images, i, m)
 
     def _get_train_loader(self, mix):
@@ -196,7 +190,6 @@ def train(self, round):
         opt = keras.optimizers.Adam(
             learning_rate=self.learning_rate, weight_decay=0.00001
         )
-        # print(self.train_loader is None)
         feature_extractor_params = self.feature_extractor.trainable_variables
         classifier_params = self.classifier.trainable_variables
         all_params = []
@@ -439,11 +432,10 @@ def evaluate(self):
         total_correct = 0
         for x, y in self.train_loader:
             logits = self.model_call(x, training=False)
-            # prob = tf.nn.softmax(logits, axis=1)
             pred = tf.argmax(logits, axis=1)
             pred = tf.cast(pred, dtype=tf.int32)
             pred = tf.reshape(pred, y.shape)
-            print(pred)
+            logging.info(pred)
             y = tf.cast(y, dtype=tf.int32)
             correct = tf.cast(tf.equal(pred, y), dtype=tf.int32)
             correct = tf.reduce_sum(correct)

examples/cifar100/fci_ssl/glfc/algorithm/data_prepocessor.py

@@ -34,7 +34,6 @@ def __init__(
         print(f"mean: {self.mean}, std: {self.std}")
 
     def preprocess_labeled_dataset(self, x, y, batch_size):
-        # wx = self.weak_augment_helper(x)
         return (
             tf.data.Dataset.from_tensor_slices((x, y))
             .shuffle(100000)
@@ -48,7 +47,6 @@ def preprocess_labeled_dataset(self, x, y, batch_size):
         )
 
     def preprocess_unlabeled_dataset(self, ux, uy, batch_size):
-        # unlabeled_train_db = tf.data.Dataset.from_tensor_slices((ux, ux, ux, uy))
 
         wux = self.weak_augment_helper(ux)
         sux = self.strong_augment_helper(ux)

examples/cifar100/fci_ssl/glfc/algorithm/proxy_server.py

@@ -157,7 +157,6 @@ def reconstruction(self):
 
         for label_i in range(100):
             if class_ratio[0][label_i] > 0:
-                # num_agumentation = self.num_image
                 agumentation = []
 
                 grad_index = np.where(proto_label == label_i)

examples/cifar100/federated_class_incremental_learning/fedavg/algorithm/network.py

@@ -27,7 +27,6 @@ def __init__(self, num_classes, feature_extractor):
         self.fc = Dense(num_classes, activation="softmax")
 
     def call(self, inputs):
-        # print(type(self.feature))
         x = self.feature(inputs)
         x = self.fc(x)
         return x

examples/cifar100/federated_class_incremental_learning/fedavg/algorithm/resnet.py

@@ -79,7 +79,6 @@ def __init__(self, layer_dims, num_classes=10):  # [2, 2, 2, 2]
 
         # output: [b, 512, h, w],
         self.avgpool = keras.layers.GlobalAveragePooling2D()
-        # self.fc = keras.layers.Dense(num_classes)
 
     def call(self, inputs, training=None):
         x = self.stem(inputs, training=training)
@@ -91,8 +90,6 @@ def call(self, inputs, training=None):
 
         # [b, c]
         x = self.avgpool(x)
-        # [b, 100]
-        # x = self.fc(x)
         return x
 
     def build_resblock(self, filter_num, blocks, stride=1):

examples/cifar100/federated_learning/fedavg/algorithm/aggregation.py

@@ -14,7 +14,6 @@
 
 import abc
 from copy import deepcopy
-from typing import List
 
 import numpy as np
 from sedna.algorithms.aggregation.aggregation import BaseAggregation
@@ -32,7 +31,7 @@ def aggregate(self, clients):
 
         Parameters
         ----------
-        clients: List
+        clients: 
             All clients in federated learning job
 
         Returns