2019-06-06 15:21:15 +03:00

41 lines
1.6 KiB
Scala

/*******************************************************************************
* Copyright (c) 2015-2018 Skymind, Inc.
*
* This program and the accompanying materials are made available under the
* terms of the Apache License, Version 2.0 which is available at
* https://www.apache.org/licenses/LICENSE-2.0.
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
* License for the specific language governing permissions and limitations
* under the License.
*
* SPDX-License-Identifier: Apache-2.0
******************************************************************************/
package org.deeplearning4j.scalnet.utils
import org.nd4j.linalg.dataset.DataSet
import org.nd4j.linalg.factory.Nd4j
/**
* A sequence generator that output toy examples for sequence classification
* Features are ramdom values between 0 and 1 and class is based on whether
* cumulative sum crossed 'timesteps * threshold'.
* ie. for 10 timesteps and 0.25 threshold: [0.6 0.2 0.9 0.9 0.3 0.6 0.8 0.1 0.8 0.2] [0 0 0 1 1 1 1 1 1 1]
*/
object SequenceGenerator {
def generate(timesteps: Int, threshold: Double = 0.25): DataSet = {
val x = Nd4j.rand(1, timesteps)
val y = Nd4j.create(1, timesteps)
for (i <- 0 until timesteps) {
val cumulativeSum = Nd4j.cumsum(x.getRow(0), 1)
val limit = timesteps * threshold
y.putScalar(0, i, if (cumulativeSum.getDouble(0l, i) > limit) 1 else 0)
}
new DataSet(x.reshape(1, timesteps, 1), y.reshape(1, timesteps, 1))
}
}