Tensorflow: Input Sparse Pipeline for SVM Evaluation

Question

Introduction:

I am trying to prepare svm evaluation of tensor flow tensorflow.contrib.learn.python.learn.estimators.svmwith sparse data. An example of use with sparse data in the github repository in tensorflow/contrib/learn/python/learn/estimators/svm_test.py#L167(I am not allowed to post more links, so here is the relative path).

The svm evaluator expects as a parameter example_id_columnand feature_columnswhere function columns should be inferred from the class FeatureColumn, for example tf.contrib.layers.feature_column.sparse_column_with_hash_bucket. See the Github Repository tensorflow/contrib/learn/python/learn/estimators/svm.py#L85and the tensorflow.org documentation for python/contrib.layers#Feature_columns.

Question:

• How do I configure my input pipeline to format sparse data so that I can use one of the tf.contrib.layers feature_columns as input to evaluate svm.
• What would a dense input function look like with many functions?

Background

The data I use is a dataset a1afrom the LIBSVM website . The data set has 123 functions (which will correspond to 123 feature_columns if the data is dense). I wrote to user op to read the type data tf.decode_csv(), but for the LIBSVM format. Op returns labels as a dense tensor, and functions as a sparse tensor. My input pipeline:

NUM_FEATURES = 123
batch_size = 200

# my op to parse the libsvm data
decode_libsvm_module = tf.load_op_library('./libsvm.so')

def input_pipeline(filename_queue, batch_size):
    with tf.name_scope('input'):
        reader = tf.TextLineReader(name="TextLineReader_")
        _, libsvm_row = reader.read(filename_queue, name="libsvm_row_")
        min_after_dequeue = 1000
        capacity = min_after_dequeue + 3 * batch_size
        batch = tf.train.shuffle_batch([libsvm_row], batch_size=batch_size,
                                       capacity=capacity,
                                       min_after_dequeue=min_after_dequeue,
                                       name="text_line_batch_")
        labels, sp_indices, sp_values, sp_shape = \
            decode_libsvm_module.decode_libsvm(records=batch,
                                               num_features=123,
                                               OUT_TYPE=tf.int64, 
                                               name="Libsvm_decoded_")
        # Return the features as sparse tensor and the labels as dense
        return tf.SparseTensor(sp_indices, sp_values, sp_shape), labels

Here's an example package with batch_size = 5.

def input_fn(dataset_name):
    maybe_download()

    filename_queue_train = tf.train.string_input_producer([dataset_name], 
                                                        name="queue_t_")
    features, labels = input_pipeline(filename_queue_train, batch_size)

    return {
        'example_id': tf.as_string(tf.range(1,123,1,dtype=tf.int64)),
        'features': features
    }, labels

This is what I have tried so far:

with tf.Session().as_default() as sess:
    sess.run(tf.global_variables_initializer())

    coord = tf.train.Coordinator()
    threads = tf.train.start_queue_runners(sess=sess, coord=coord)

    feature_column = tf.contrib.layers.sparse_column_with_hash_bucket(
        'features', hash_bucket_size=1000, dtype=tf.int64)

    svm_classifier = svm.SVM(feature_columns=[feature_column],
                             example_id_column='example_id',
                             l1_regularization=0.0,
                             l2_regularization=1.0)
    svm_classifier.fit(input_fn=lambda: input_fn(TRAIN),
                       steps=30)

    accuracy = svm_classifier.evaluate(
        input_fn= lambda: input_fn(features, labels), 
        steps=1)['accuracy']                       
    print(accuracy)
    coord.request_stop()

    coord.join(threads)
    sess.close()

Answer 1

, , TensorFlow 1.1.0-rc2. , ; ~ 100 (tf.sparse_tensor_to_dense) real_valued_column, sparse_column_with_integerized_feature SVM.

import tensorflow as tf

batch_size = 10
num_features = 123
num_examples = 100

def input_fn():
  example_ids = tf.random_uniform(
      [batch_size], maxval=num_examples, dtype=tf.int64)
  # Construct a SparseTensor with features
  dense_features = (example_ids[:, None]
                    + tf.range(num_features, dtype=tf.int64)[None, :]) % 2
  non_zeros = tf.where(tf.not_equal(dense_features, 0))
  sparse_features = tf.SparseTensor(
      indices=non_zeros,
      values=tf.gather_nd(dense_features, non_zeros),
      dense_shape=[batch_size, num_features])
  features = {
      'some_sparse_features': tf.sparse_tensor_to_dense(sparse_features),
      'example_id': tf.as_string(example_ids)}
  labels = tf.equal(dense_features[:, 0], 1)
  return features, labels
svm = tf.contrib.learn.SVM(
    example_id_column='example_id',
    feature_columns=[
      tf.contrib.layers.real_valued_column(
          'some_sparse_features')],
    l2_regularization=0.1, l1_regularization=0.5)
svm.fit(input_fn=input_fn, steps=1000)
positive_example = lambda: {
    'some_sparse_features': tf.sparse_tensor_to_dense(
        tf.SparseTensor([[0, 0]], [1], [1, num_features])),
    'example_id': ['a']}
print(svm.evaluate(input_fn=input_fn, steps=20))
print(next(svm.predict(input_fn=positive_example)))
negative_example = lambda: {
    'some_sparse_features': tf.sparse_tensor_to_dense(
        tf.SparseTensor([[0, 0]], [0], [1, num_features])),
    'example_id': ['b']}
print(next(svm.predict(input_fn=negative_example)))

{'accuracy': 1.0, 'global_step': 1000, 'loss': 1.0645389e-06}
{'logits': array([ 0.01612902], dtype=float32), 'classes': 1}
{'logits': array([ 0.], dtype=float32), 'classes': 0}