data.h

Go to the documentation of this file.

// Copyright (c) 2021, Aalto University, developed by Erik Schultheis
// All rights reserved.
//
// SPDX-License-Identifier: MIT
 
#ifndef DISMEC_DATA_H
#define DISMEC_DATA_H
 
#include <memory>
#include "matrix_types.h"
#include "data/types.h"
#include "utils/eigen_generic.h"
 
namespace dismec {
    class DatasetBase {
    public:
        virtual ~DatasetBase() = default;
        DatasetBase(const DatasetBase&) = default;
        DatasetBase(DatasetBase&&) = default;
        DatasetBase& operator=(DatasetBase&&) = default;
        DatasetBase& operator=(const DatasetBase&) = default;
 
        [[nodiscard]] std::shared_ptr<const GenericFeatureMatrix> get_features() const;
 
        [[nodiscard]] std::shared_ptr<GenericFeatureMatrix> edit_features();
 
        [[nodiscard]] long num_features() const noexcept;
 
        [[nodiscard]] long num_examples() const noexcept;
 
        [[nodiscard]] virtual long num_labels() const noexcept = 0;
 
        [[nodiscard]] virtual long num_positives(label_id_t id) const;
 
        [[nodiscard]] virtual long num_negatives(label_id_t id) const;
 
        [[nodiscard]] std::shared_ptr<const BinaryLabelVector> get_labels(label_id_t id) const;
 
        virtual void get_labels(label_id_t id, Eigen::Ref<BinaryLabelVector> target) const = 0;
    protected:
        explicit DatasetBase(SparseFeatures x);
        explicit DatasetBase(DenseFeatures x);
 
        // features
        std::shared_ptr<GenericFeatureMatrix> m_Features;
    };
 
    class BinaryData : public DatasetBase {
    public:
        BinaryData(SparseFeatures x, std::shared_ptr<BinaryLabelVector> y) :
                DatasetBase(std::move(x)), m_Labels(std::move(y))
        {
 
        }
 
        [[nodiscard]] long num_labels() const noexcept override;
        void get_labels(label_id_t i, Eigen::Ref<BinaryLabelVector> target) const override;
    private:
 
        // targets
        std::shared_ptr<BinaryLabelVector> m_Labels;
    };
 
 
    class MultiLabelData : public DatasetBase {
    public:
        MultiLabelData(SparseFeatures x, std::vector<std::vector<long>> y) :
                DatasetBase(x.markAsRValue()), m_Labels(std::move(y)) {
 
        }
 
        MultiLabelData(DenseFeatures x, std::vector<std::vector<long>> y) :
            DatasetBase(std::move(x)), m_Labels(std::move(y)) {
        }
 
        [[nodiscard]] long num_labels() const noexcept override;
        void get_labels(label_id_t label, Eigen::Ref<BinaryLabelVector> target) const override;
 
        // these are faster than the default implementation
        [[nodiscard]] long num_positives(label_id_t id) const override;
        [[nodiscard]] long num_negatives(label_id_t id) const override;
 
        [[nodiscard]] const std::vector<long>& get_label_instances(label_id_t label) const;
 
        void select_labels(label_id_t start, label_id_t end);
 
        [[nodiscard]] const std::vector<std::vector<long>>& all_labels() const { return m_Labels; }
    private:
        // targets: vector of vectors of example ids: if label i is present in example j, `j \in then m_Labels[i]`
        std::vector<std::vector<long>> m_Labels;
    };
}
 
#endif //DISMEC_DATA_H