1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120

//! Simple interfaces to import data.
use std::fmt;
use std::io;

use crate::tensor::*;

pub(crate) use self::batch_iterator::BatchIterator;
pub use self::image_data::ImageDataSet;
pub use self::image_data::ImageDataSetBuilder;
pub use self::image_data::ImageOps;
pub use self::tabular_data::TabularDataSet;

mod batch_iterator;
mod image_data;
mod tabular_data;

/// Errors that may be raised by data sets methods.
#[derive(Debug)]
pub enum DataSetError {
    Io(io::Error),
    Csv(csv::Error),
    DimensionMismatch,
    PathDoesNotExist,
    TrainPathDoesNotExist,
    ValidPathDoesNotExist,
    ImageFormatNotSupported,
    InvalidImagePath,
    InvalidValidationFraction,
    DifferentNumbersOfChannels,
}

/// Types of data.
#[derive(Debug)]
enum IO {
    Input,
    Output,
}

impl fmt::Display for DataSetError {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        match *self {
            DataSetError::Io(ref err) => write!(f, "IO error: {}", err),
            DataSetError::Csv(ref err) => write!(f, "CSV error: {}", err),
            DataSetError::DimensionMismatch => write!(f, "The number of input and output samples differ."),
            DataSetError::PathDoesNotExist => write!(f, "The path does not exist."),
            DataSetError::TrainPathDoesNotExist => write!(f, "The root directory does not contain a 'train' subfolder."),
            DataSetError::ValidPathDoesNotExist => write!(f, "The root directory does not contain a 'valid' subfolder."),
            DataSetError::ImageFormatNotSupported => write!(f, "The image format is not supported."),
            DataSetError::InvalidImagePath => write!(f, "The path could not be opened as an image."),
            DataSetError::InvalidValidationFraction => write!(f, "The validation fraction is incorrect. It must be between 0 and 1."),
            DataSetError::DifferentNumbersOfChannels => write!(f, "The directory contains images with different numbers of channels."),
        }
    }
}

impl std::convert::From<io::Error> for DataSetError {
    fn from(error: io::Error) -> DataSetError {
        DataSetError::Io(error)
    }
}

/// Defines the type of scaling that has been performed.
#[derive(Debug)]
pub enum Scaling {
    Normalized,
    Standarized,
}


/// Trait that must be implemented for any type of dataset supported by neuro.
pub trait DataSet {
    /// Returns the dimension of the samples.
    fn input_shape(&self) -> Dim;

    /// Returns the dimension of the labels.
    fn output_shape(&self) -> Dim;

    /// Returns the number of samples in the training set.
    fn num_train_samples(&self) -> u64;

    /// Returns the number of samples in the validation set.
    fn num_valid_samples(&self) -> u64;

    /// Returns the classes in the data set.
    fn classes(&self) -> Option<Vec<String>> { None }

    /// Returns a reference to the training samples.
    fn x_train(&self) -> &Tensor;

    /// Returns a reference to the training labels.
    fn y_train(&self) -> &Tensor;

    /// Returns a reference to the validation samples.
    fn x_valid(&self) -> Option<&Tensor>;

    /// Returns a reference to the validation labels.
    fn y_valid(&self) -> Option<&Tensor>;

    /// Returns a reference to the test samples.
    fn x_test(&self) -> Option<&Tensor>;

    /// Returns a reference to the test labels.
    fn y_test(&self) -> Option<&Tensor>;

    /// Returns a reference to the type of scaling that has been applied to the input features and the values used for the scaling.
    ///
    /// If scaling has been applied, the returned tuple contains the following:
    /// * Normalization: (Scaling::Normalized, minimum value, maximum value)
    /// * Standardization: (Scaling::Standarized, mean, standard deviation)
    ///
    fn x_train_stats(&self) -> &Option<(Scaling, Tensor, Tensor)>;

    /// Returns a reference to the type of scaling that has been applied to the output labels and the values used for the scaling.
    ///
    /// If scaling has been applied, the returned tuple contains the following:
    /// * Normalization: (Scaling::Normalized, minimum value, maximum value)
    /// * Standardization: (Scaling::Standarized, mean, standard deviation)
    ///
    fn y_train_stats(&self) -> &Option<(Scaling, Tensor, Tensor)>;
}