def train_test_split(unrolled, test_size):
cutPoint = unrolled.count() - test_size
train = unrolled.filter(lambda x: x.index < cutPoint) \
.map(lambda x: Sample.from_ndarray(np.array(x.feature), np.array(x.label)))
test = unrolled.filter(lambda x: x.index >= cutPoint) \
.map(lambda x: Sample.from_ndarray(np.array(x.feature), np.array(x.label)))
return [train, test]
def gen_rand_user_item_feature(user_num, item_num, class_num):
user_id = random.randint(1, user_num)
item_id = random.randint(1, item_num)
rating = random.randint(1, class_num)
sample = Sample.from_ndarray(np.array([user_id, item_id]),
np.array([rating]))
return UserItemFeature(user_id, item_id, sample)
def to_sample(vectors, label, embedding_dim):
# flatten nested list
flatten_features = list(itertools.chain(*vectors))
features = np.array(flatten_features, dtype='float').reshape(
[sequence_len, embedding_dim])
return Sample.from_ndarray(features, np.array(label))
def to_sample(vectors, label, embedding_dim):
# flatten nested list
flatten_features = list(itertools.chain(*vectors))
features = np.array(flatten_features, dtype='float').reshape(
[sequence_len, embedding_dim])
return Sample.from_ndarray(features, np.array(label))
def get_featureset(x, y, shuffle=True):
x = np.split(x.data.numpy(), x.shape[0])
y = np.split(y.data.numpy(), y.shape[0])
print(x[0].shape)
print(y[0].shape)
samples = [Sample.from_ndarray(np.squeeze(x[i]), np.squeeze(y[i])) for i in range(len(x))]
sample_rdd = sc.parallelize(samples)
return FeatureSet.sample_rdd(sample_rdd, shuffle=shuffle)
def to_sample(vectors, label, embedding_dim):
# flatten nested list
flatten_features = list(itertools.chain(*vectors))
features = np.array(flatten_features,
dtype='float').reshape([sequence_len, embedding_dim])
if model_type.lower() == "cnn":
features = features.transpose(1, 0)
return Sample.from_ndarray(features, np.array(label))
def to_sample(data):
from bigdl.util.common import Sample
data = check_type_and_convert(data, allow_list=True, allow_tuple=False)
features = data["x"]
labels = data["y"]
length = features[0].shape[0]
for i in range(length):
fs = [feat[i] for feat in features]
ls = [l[i] for l in labels]
yield Sample.from_ndarray(np.array(fs), np.array(ls))
def to_sample_rdd(x, y, numSlices=None):
"""
Conver x and y into RDD[Sample]
:param x: ndarray and the first dimension should be batch
:param y: ndarray and the first dimension should be batch
:param numSlices:
:return:
"""
from bigdl.util.common import Sample
x_rdd = sc.parallelize(x, numSlices)
y_rdd = sc.parallelize(y, numSlices)
return x_rdd.zip(y_rdd).map(lambda item: Sample.from_ndarray(item[0], item[1]))
def row_to_sample(row, column_info, model_type="wide_n_deep"):
wide_tensor = get_wide_tensor(row, column_info)
deep_tensor = JTensor.from_ndarray(get_deep_tensor(row, column_info))
label = row[column_info.label]
model_type = model_type.lower()
if model_type == "wide_n_deep":
feature = [wide_tensor, deep_tensor]
elif model_type == "wide":
feature = wide_tensor
elif model_type == "deep":
feature = deep_tensor
else:
raise TypeError("Unsupported model_type: %s" % model_type)
return Sample.from_jtensor(feature, label)
def row_to_sample(row, column_info, model_type="wide_n_deep"):
wide_tensor = get_wide_tensor(row, column_info)
deep_tensor = JTensor.from_ndarray(get_deep_tensor(row, column_info))
label = row[column_info.label]
model_type = model_type.lower()
if model_type == "wide_n_deep":
feature = [wide_tensor, deep_tensor]
elif model_type == "wide":
feature = wide_tensor
elif model_type == "deep":
feature = deep_tensor
else:
raise TypeError("Unsupported model_type: %s" % model_type)
return Sample.from_jtensor(feature, label)
def row_to_sample(row, column_info, model_type="wide_n_deep"):
"""
convert a row to sample given column feature information of a WideAndDeep model
:param row: Row of userId, itemId, features and label
:param column_info: ColumnFeatureInfo specify information of different features
:return: TensorSample as input for WideAndDeep model
"""
wide_tensor = get_wide_tensor(row, column_info)
deep_tensor = get_deep_tensors(row, column_info)
deep_tensors = [JTensor.from_ndarray(ele) for ele in deep_tensor]
label = row[column_info.label]
model_type = model_type.lower()
if model_type == "wide_n_deep":
feature = [wide_tensor] + deep_tensors
elif model_type == "wide":
feature = wide_tensor
elif model_type == "deep":
feature = deep_tensors
else:
raise TypeError("Unsupported model_type: %s" % model_type)
return Sample.from_jtensor(feature, label)
import matplotlib.pyplot as plt
from pyspark import SparkContext
from bigdl.util.common import init_engine
from bigdl.util.common import create_spark_conf
from bigdl.util.common import JavaCreator
from bigdl.util.common import Sample
from vision.image3d.transformation import *
import h5py
from math import pi
img_path = os.path.abspath(__file__ + "/../../resources/image_3d/a.mat")
sample = h5py.File(img_path)['meniscus_im']
sample = np.array(sample)
sample = Sample.from_ndarray(features=sample, label=np.array(-1))
# sample = np.expand_dims(sample,0)
print(sample.features[0].shape)
sc = SparkContext(appName="test", conf=create_spark_conf())
JavaCreator.set_creator_class(
"com.intel.analytics.zoo.transform.vision.image3d.python.api.VisionPythonBigDL"
)
init_engine()
data_rdd = sc.parallelize([sample])
start_loc = [13, 80, 125]
patch = [5, 40, 40]
# end_loc = [17,119,164]
def gen_rand_user_item_feature(user_num, item_num, class_num):
user_id = random.randint(1, user_num)
item_id = random.randint(1, item_num)
rating = random.randint(1, class_num)
sample = Sample.from_ndarray(np.array([user_id, item_id]), np.array([rating]))
return UserItemFeature(user_id, item_id, sample)
def prepare_data(sc, folder, vocabsize, training_split):
if not folder.startswith('hdfs://'):
file = download_data(folder)
else:
file = folder
sentences_rdd = sc.textFile(file) \
.map(lambda line: sentence.sentences_split(line))
pad_sent = sentences_rdd.flatMap(lambda x: x). \
map(lambda sent: sentence.sentences_bipadding(sent))
tokens = pad_sent.map(lambda pad: sentence.sentence_tokenizer(pad))
train_tokens, val_tokens = tokens.randomSplit(
[training_split, 1 - training_split])
train_tokens.cache()
val_tokens.cache()
train_max_len = train_tokens.map(lambda x: len(x)).max()
print("max length %s" % train_max_len)
words = train_tokens.flatMap(lambda x: x)
print("%s words and %s sentences processed in train data" %
(words.count(), train_tokens.count()))
val_max_len = val_tokens.map(lambda x: len(x)).max()
print("val max length %s" % val_max_len)
val_words = val_tokens.flatMap(lambda x: x)
print("%s words and %s sentences processed in validation data" %
(val_words.count(), val_tokens.count()))
sort_words = words.map(lambda w: (w, 1)) \
.reduceByKey(lambda a, b: a + b) \
.sortBy(lambda w_c: w_c[1])
vocabulary = np.array(sort_words.map(lambda w: w[0]).collect())
fre_len = vocabulary.size
if vocabsize > fre_len:
length = fre_len
else:
length = vocabsize
discard_vocab = vocabulary[:fre_len - length]
used_vocab = vocabulary[fre_len - length:fre_len]
used_vocab_size = used_vocab.size
index = np.arange(used_vocab_size)
index2word = dict(enumerate(used_vocab))
word2index = dict(zip(used_vocab, index))
total_vocab_len = used_vocab_size + 1
startIdx = word2index.get("SENTENCESTART")
endIdx = word2index.get("SENTENCEEND")
def text2labeled(sent):
indexes = [word2index.get(x, used_vocab_size) for x in sent]
data = indexes[0:-1]
label = indexes[1:len(indexes)]
return data, label
def labeled2onehotformat(labeled_sent):
label = [x + 1 for x in labeled_sent[1]]
size = len(labeled_sent[0])
feature_onehot = np.zeros(size * total_vocab_len,
dtype='int').reshape([size, total_vocab_len])
for i, el in enumerate(labeled_sent[0]):
feature_onehot[i, el] = 1
return feature_onehot, label
def padding(features, label, length):
pad_len = length - len(label)
padded_label = (label + [startIdx] * length)[:length]
feature_padding = np.zeros((pad_len, total_vocab_len), dtype=np.int)
feature_padding[:, endIdx + 1] = np.ones(pad_len)
padded_feautres = np.concatenate((features, feature_padding), axis=0)
return padded_feautres, padded_label
sample_rdd = train_tokens.map(lambda sentence_te: text2labeled(sentence_te)) \
.map(lambda labeled_sent: labeled2onehotformat(labeled_sent)) \
.map(lambda x: padding(x[0], x[1], train_max_len)) \
.map(lambda vectors_label: Sample.from_ndarray(vectors_label[0],
np.array(vectors_label[1]))).cache()
val_sample_rdd = val_tokens.map(lambda sentence_t: text2labeled(sentence_t)) \
.map(lambda labeled_sent: labeled2onehotformat(labeled_sent)) \
.map(lambda x: padding(x[0], x[1], val_max_len)) \
.map(lambda vectors_label: Sample.from_ndarray(vectors_label[0],
np.array(vectors_label[1]))).cache()
return sample_rdd, val_sample_rdd, total_vocab_len
def prepare_data(sc, folder, vocabsize, training_split):
if not folder.startswith( 'hdfs://' ):
file = download_data(folder)
else:
file = folder
sentences_rdd = sc.textFile(file) \
.map(lambda line: sentence.sentences_split(line))
pad_sent = sentences_rdd.flatMap(lambda x: x). \
map(lambda sent: sentence.sentences_bipadding(sent))
tokens = pad_sent.map(lambda pad: sentence.sentence_tokenizer(pad))
train_tokens, val_tokens = tokens.randomSplit([training_split, 1 - training_split])
train_tokens.cache()
val_tokens.cache()
train_max_len = train_tokens.map(lambda x: len(x)).max()
print("max length %s" % train_max_len)
words = train_tokens.flatMap(lambda x: x)
print("%s words and %s sentences processed in train data" % (words.count(), train_tokens.count()))
val_max_len = val_tokens.map(lambda x: len(x)).max()
print("val max length %s" % val_max_len)
val_words = val_tokens.flatMap(lambda x: x)
print("%s words and %s sentences processed in validation data" % (val_words.count(), val_tokens.count()))
sort_words = words.map(lambda w: (w, 1)) \
.reduceByKey(lambda a, b: a + b) \
.sortBy(lambda w_c: w_c[1])
vocabulary = np.array(sort_words.map(lambda w: w[0]).collect())
fre_len = vocabulary.size
if vocabsize > fre_len:
length = fre_len
else:
length = vocabsize
discard_vocab = vocabulary[: fre_len-length]
used_vocab = vocabulary[fre_len-length: fre_len]
used_vocab_size = used_vocab.size
index = np.arange(used_vocab_size)
index2word = dict(enumerate(used_vocab))
word2index = dict(zip(used_vocab, index))
total_vocab_len = used_vocab_size + 1
startIdx = word2index.get("SENTENCESTART")
endIdx = word2index.get("SENTENCEEND")
def text2labeled(sent):
indexes = [word2index.get(x, used_vocab_size) for x in sent]
data = indexes[0: -1]
label = indexes[1: len(indexes)]
return data, label
def labeled2onehotformat(labeled_sent):
label = [x+1 for x in labeled_sent[1]]
size = len(labeled_sent[0])
feature_onehot = np.zeros(size * total_vocab_len, dtype='int').reshape(
[size, total_vocab_len])
for i, el in enumerate(labeled_sent[0]):
feature_onehot[i, el] = 1
return feature_onehot, label
def padding(features, label, length):
pad_len = length - len(label)
padded_label = (label + [startIdx] * length)[:length]
feature_padding = np.zeros((pad_len, total_vocab_len), dtype=np.int)
feature_padding[:, endIdx + 1] = np.ones(pad_len)
padded_feautres = np.concatenate((features, feature_padding), axis=0)
return padded_feautres, padded_label
sample_rdd = train_tokens.map(lambda sentence_te: text2labeled(sentence_te)) \
.map(lambda labeled_sent: labeled2onehotformat(labeled_sent)) \
.map(lambda x: padding(x[0], x[1], train_max_len)) \
.map(lambda vectors_label: Sample.from_ndarray(vectors_label[0],
np.array(vectors_label[1]))).cache()
val_sample_rdd = val_tokens.map(lambda sentence_t: text2labeled(sentence_t)) \
.map(lambda labeled_sent: labeled2onehotformat(labeled_sent)) \
.map(lambda x: padding(x[0], x[1], val_max_len)) \
.map(lambda vectors_label: Sample.from_ndarray(vectors_label[0],
np.array(vectors_label[1]))).cache()
return sample_rdd, val_sample_rdd, total_vocab_len
def get_rdd(x, y, shuffle=False):
x = np.split(x.data.numpy(), x.shape[0])
y = np.split(x.data.numpy(), y.shape[0])
samples = [Sample.from_ndarray(np.squeeze(x[i]), np.squeeze(y[i])) for i in range(len(x))]
sample_rdd = sc.parallelize(samples)
return sample_rdd
def normalizer(mean, std):
"""
Normalize features by standard deviation
"""
return lambda sample: Sample.from_ndarray((sample.features - mean) / std,
sample.label, sample.bigdl_type)
