scripts/nn/examples/mnist_lenet-predict.dml

#-------------------------------------------------------------
#
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements.  See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership.  The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License.  You may obtain a copy of the License at
#
#   http://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.
#
#-------------------------------------------------------------

# MNIST LeNet - Predict
#
# This script computes the class probability predictions of a
# trained convolutional net using the "LeNet" architecture on
# images of handwritten digits.
#
# Inputs:
#  - X: File containing training images.
#     The format is "pixel_1, pixel_2, ..., pixel_n".
#  - C: Number of color chanels in the images.
#  - Hin: Input image height.
#  - Win: Input image width.
#  - model_dir: Directory containing the trained weights and biases
#     of the model.
#  - out_dir: Directory to store class probability predictions for
#     each image.
#  - fmt: [DEFAULT: "csv"] File format of `X` and output predictions.
#     Options include: "csv", "mm", "text", and "binary".
#
# Outputs:
#  - probs: File containing class probability predictions for each
#     image.
#
# Data:
# The X file should contain images of handwritten digits,
# where each example is a 28x28 pixel image of grayscale values in
# the range [0,255] stretched out as 784 pixels.
#
# Sample Invocation (running from outside the `nn` folder):
# 1. Download images.
#
#   For example, save images to `nn/examples/data/mnist/images.csv`.
#
# 2. Execute using Spark
#   ```
#   spark-submit --master local[*] --driver-memory 5G
#   --conf spark.driver.maxResultSize=0 --conf spark.rpc.message.maxSize=128
#   $SYSTEMDS_ROOT/target/SystemDS.jar -f nn/examples/mnist_lenet-predict.dml
#   -nvargs X=nn/examples/data/mnist/images.csv C=1 Hin=28 Win=28
#   model_dir=nn/examples/model/mnist_lenet out_dir=nn/examples/data/mnist
#   ```
#
source("nn/examples/mnist_lenet.dml") as mnist_lenet

# Read training data
fmt = ifdef($fmt, "csv")
X = read($X, format=fmt)
C = $C
Hin = $Hin
Win = $Win

# Scale images to [-1,1]
X = (X / 255.0) * 2 - 1

# Read model coefficients
W1 = read($model_dir+"/W1")
b1 = read($model_dir+"/b1")
W2 = read($model_dir+"/W2")
b2 = read($model_dir+"/b2")
W3 = read($model_dir+"/W3")
b3 = read($model_dir+"/b3")
W4 = read($model_dir+"/W4")
b4 = read($model_dir+"/b4")

# Predict classes
probs = mnist_lenet::predict(X, C, Hin, Win, W1, b1, W2, b2, W3, b3, W4, b4)

# Output results
write(probs, $out_dir+"/probs."+fmt, format=fmt)