Intelligent Big Multimedia Databases

The following sections are included:

• Dedication
• Preface
• Contents

Multimedia databases are employed in an increasing number of commercially important applications, such as media-on-demand, surveillance systems and medical systems. Multimedia databases divide a subject into numerous major areas. Because each area is extensive, a major understanding of the mathematical core concepts requires an investigation of the different areas. In this book, we attempt to unify the essential concepts and recent algorithms.

We begin with a short introduction to relational databases and introduce examples of popular multimedia information. Multimedia data enables new data access methods, such as query by images, in which the most similar image to the presented image is determined; it is also referred to as content based image retrieval (CBIR).

Transform functions can be used for lossy and lossless compression; they are the basis of feature extraction and of high dimensional indexing techniques.

Multimedia objects are represented by digital information (numbers, vectors). The size of a multimedia object may be large; a high-quality colored image requires 6 MB and five minutes of a high-quality video clip with 30 FPS would require 54 GB. Compression is a clever method for encoding information using fewer bits than required by the original representation. Two categories of compression exist: lossless compression and lossy compression. Both types of compression can reduce the amount of the source information. Although no information is lost in lossless compression, this is not the case with lossy compression. When the compressed information is decompressed in lossy compression, a minor loss of information and quality occurs from the identification of unnecessary or unimportant information, which can be removed.

The extraction of primitive out of media data is referred to as feature extraction. The set of features represents the relevant information about the input data in a certain context. The context is dependent on the desired task that uses the reduced representation instead of the original input. The set of primitives is usually described by a feature vector. Feature extraction is related to compression algorithms and is frequently based on transform functions, as described in the previous chapter. During content-based media retrieval, the feature vectors are used to determine the similarities among the media objects. The extracted features represent the signature of the media object. Feature extraction is a crucial step in content-based media retrieval.

For the fast access of large data, divide and conquer methods that are based on hierarchical structures are employed. For numbers, a tree can be utilized to prune branches in the processing queries. The access is fast: it is logarithmic in relation to the size of the database that represents the numbers. For low-dimensional vectors, metric index trees, such as kd-trees and R-trees can be employed. Alternatively, an index structure that is based on space-filling curves can be constructed.

The growth in the number of dimensions has negative implications in the performance of multidimensional index trees. These negative effects are named as the “curse of dimensionality.” A nearest-neighbor query in a high dimensional space corresponds to a hyper-sphere with a huge radius which is mostly larger than the extension of the data space in most dimensions…

Traditional indexing of multimedia data leads to dilemma. Either the number of features has to be reduced or the quality of the results in unsatisfactory, or approximate queries is preformed leading to a relative error during retrieval. The promise of the recently introduced subspace-tree is the logarithmic retrieval complexity of extremely high dimensional features. The subspace-tree indicates that the conjecture “the curse of dimensionality” could be false. The search in such a structure starts at the subspace with the lowest dimension. In this subspace, the set of all possible similar objects is determined. In the next subspace, additional metric information corresponding to a higher dimension is used to reduce this set. This process is then repeated. The theoretical estimation of temporal complexity of the subspace tree is logarithmic for the Gaussian (Normal) distribution of the distances between the data points.

The descriptor represents the relevant information about a text by a feature vector that indicates the presence or absence of terms. Terms are words with specific meaning in a specific context; they may deviate from the meaning the same words have in other contexts. The frequency of the occurrence of each term in a document and the information content of the term according to the entire document collection can be employed. During information retrieval, the feature vectors are used to determine the similarity between two text documents, which is represented by the cosine of the angle between the feature vectors. Because the vectors are extremely sparse (most dimensions are zero), we cannot apply the previously introduced indexing methods. The semantic gap exists because the text documents are described by feature vectors without additional semantical information. Text understanding systems are developed in the field of AI; however, they only work in a narrow domain and are computationally expensive.

Several parallels between human learning and machine learning exist. Some of the techniques are inspired from the efforts of psychologists and biologists to simulate human learning via computational models. In addition to statistical machine learning, other forms of machine learning, such as inductive learning, knowledge learning, analogical learning and reinforcement learning, exist but will not be addressed in this chapter.

A multimodal search enables an information search using search queries in multiple data types, including text and other multimedia formats. The information is described by some feature vectors and categories that were determined by indexing structures or supervised learning algorithms. A feature vector or category can belong to different modalities, such as word, shape, and color.

We highlight some basic architecture issues related to multimedia databases and big data. Big data is a large collection of unstructured data that cannot be processed with traditional methods, such as standard database management systems.

We present multimedia examples of database applications in medicine. A hospital information system (HIS) is a comprehensive and integrated information system that is designed to manage the administrative, financial and clinical aspects of a hospital. A clinical information system (CIS) is frequently separated from a HIS due to its focus on patient-state- and clinical-state-related data. A clinical information system can be composed of a variety of subsystems in medical specialties. An example of a clinical information system is the electronic patient record (EHR). It includes information related to the current and historical health, medical conditions and medical imaging.

The following sections are included:

• Bibliography
• Index