Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

Data Mining - Concepts and Techniques.

Knowledge Management

Machine learning (ML) is the fastest growing field in computer science, and health informatics is among the greatest challenges. The goal of ML is to develop algorithms which can learn and improve over time and can be used for predictions. Most ML researchers concentrate on automatic machine learning (aML), where great advances have been made, for example, in speech recognition, recommender systems, or autonomous vehicles. Automatic approaches greatly benefit from big data with many training sets. However, in the health domain, sometimes we are confronted with a small number of data sets or rare events, where aML-approaches suffer of insufficient training samples. Here interactive machine learning (iML) may be of help, having its roots in reinforcement learning, preference learning, and active learning. The term iML is not yet well used, so we define it as “algorithms that can interact with agents and can optimize their learning behavior through these interactions, where the agents can also be human.” This “human-in-the-loop” can be beneficial in solving computationally hard problems, e.g., subspace clustering, protein folding, or k-anonymization of health data, where human expertise can help to reduce an exponential search space through heuristic selection of samples. Therefore, what would otherwise be an NP-hard problem, reduces greatly in complexity through the input and the assistance of a human agent involved in the learning phase.

/pdf/interactive-machine-learning-for-health-informatics-when-do-434i9eplg1.pdf

Interactive machine learning for health informatics: when do we need the human-in-the-loop?

Photoinhibition of Photosystem II (PSII) is the light-induced loss of PSII electron-transfer activity. Although photoinhibition has been studied for a long time, there is no consensus about its mechanism. On one hand, production of singlet oxygen ((1)O(2)) by PSII has promoted models in which this reactive oxygen species (ROS) is considered to act as the agent of photoinhibitory damage. These chemistry-based models have often not taken into account the photophysical features of photoinhibition-like light response and action spectrum. On the other hand, models that reproduce these basic photophysical features of the reaction have not considered the importance of data about ROS. In this chapter, it is shown that the evidence behind the chemistry-based models and the photophysically oriented models can be brought together to build a mechanism that confirms with all types of experimental data. A working hypothesis is proposed, starting with inhibition of the manganese complex by light. Inability of the manganese complex to reduce the primary donor promotes recombination between the oxidized primary donor and Q(A), the first stable quinone acceptor of PSII. (1)O(2) production due to this recombination may inhibit protein synthesis or spread the photoinhibitory damage to another PSII center. The production of (1)O(2) is transient because loss of activity of the oxygen-evolving complex induces an increase in the redox potential of Q(A), which lowers (1)O(2) production.

Photoinhibition of Photosystem II

Semantic Search emerged as the new system paradigm in enterprise information systems. However, usually only small amounts of textual enterprise data is semantically prepared for such systems. The manual semantification of these resources typically is a time-consuming process. The automatic semantification requires deep knowledge in Natural Language Processing. Therefore, in this paper we present a novel approach that makes the underlying Subject Indexing task rather a Knowledge Engineering than a Natural Language Processing task. The approach is based on a simple but powerful and intuitive probabilistic model that allows for the easy integration of expert knowledge.

/pdf/the-revieval-of-subject-analysis-a-knowledge-based-approach-1j5h85g8je.pdf

The Revieval of Subject Analysis: A Knowledge-based Approach facilitating Semantic Search.

The success of knowledge systems for diagnostic tasks has been proved in the last decades. Here, the evaluation of the developed knowledge system is an important and critical issue to deal with. Besides analysis techniques, such as manual inspection and static verification the use of empirical testing (running previously solved test cases) is the most common approach for the validation of knowledge systems. In the past, some approaches for the automated generation of test cases have been presented, cf. (Gupta & Biegel 1990; Gonzalez & Dankel 1993; Knauf, Gonzalez, & Abel 2002; Knauf et al. 2004). Such conservative methods are useful for generating suitable test cases in arbitrary domains, but require the availability of explicit knowledge, either represented as already formalized knowledge (e.g. rules) or described as generation knowledge (e.g. constraints or causal dependency models). In this paper, we briefly introduce a novel method for test case generation which is appropriate in the context of a testfirst approach. Test-first approaches postulate the development of suitable test cases before the corresponding functionality is added to the system, e.g. knowledge is formalized and added to the knowledge base. Test-first approaches, e.g. (Baumeister, Seipel, & Puppe 2004) have some advantages when compared to conservative approaches: 1) The developer needs to think about the desired functionality of the system before coding knowledge. 2) Test knowledge is not acquired depended on particular knowledge slices, e.g. rules. 3) Unexpected errors (compared to conservative approaches) in knowledge can be detected due to the creative nature of the approach, e.g. cases that were defined to derive a particular solution but actually do not.

/pdf/semi-automatic-generation-of-test-cases-by-case-morphing-1ps1a2w94d.pdf

Semi-automatic generation of test cases by case morphing

In this paper we present a visual representation of temporal patterns in abstractions of numerical and time-stamped data We provide a curve-like acquisition tool which supports domain specialists to develop and refine temporal knowledge in an intuitive and effective manner The resulting patterns can be used to detect artifacts as well as more complex phenomena, eg, in order to derive intelligent alarms Some experiences obtained from the acquisition and the use of medical surgery patterns is reported

/pdf/relations-between-visual-and-textual-representations-of-x16932ylop.pdf

Relations between Visual and Textual Representations of Temporal Patterns for Medical Data Abstraction

Recently, semantic technologies improved the information access to massive amounts of data in various manners. Also industrial domains benefit from semantic approaches. Information now can be accessed more precisely and is directly linked with related data. However, the actual support for humans to access suitable content at time is still in its infancy. In this paper, we motivate a number of requirements for extending current state-of-the-art information systems in order to support human-aware information access. The requirements are based on the experiences we made during the introduction of industrial information systems over the last years.

/pdf/requirements-of-affective-information-systems-in-the-uiasgftkvb.pdf

Requirements of Affective Information Systems in the Industrial Domain.

Current industrial information systems maintain and process immense amounts of knowledge to be used for the development, configuration, and production of artifacts. From a logical point of view these elements process knowledge abozt the same concepts albeit existing in different information systems. From a practical point of view, however, the systems are only loosely coupled and only have little linkage. These semantic gaps prevent a seamless analysis and evaluation of the entire knowledge, which in practice, is of significant importance for productional success. In this paper, we propose a general high-level ontology that integrates with the knowledge of the information systems and enables the analysis and evaluation tasks. We describe the basic concepts and properties of industrial knowledge and motivate a number of general evaluation tasks.

/pdf/evaluation-of-configuration-knowledge-in-industrial-2aej80knmj.pdf

Joachim Baumeister

Papers

The Revieval of Subject Analysis: A Knowledge-based Approach facilitating Semantic Search.

Semi-automatic generation of test cases by case morphing

Relations between Visual and Textual Representations of Temporal Patterns for Medical Data Abstraction

Requirements of Affective Information Systems in the Industrial Domain.

Evaluation of Configuration Knowledge in Industrial Manufacturing.