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

A prevalent and pervasive view of designing is that it can be modeled using variables and decisions made about what values should be taken by these variables. The activity of designing is carried out with the expectation that the designed artifact will operate in the natural world and the social world. These worlds impose constraints on the variables and their values; so, design could be described as a goal-oriented, constrained, decision- making activity. However, design distinguish- es itself from other similarly described activities not only by its domain but also by additional necessary features. Designing involves exploration, exploring what variables might be appropriate. The process of explo- ration involves both goal variables and deci- sion variables. In addition, designing involves learning: Part of the exploration activity is learning about emerging features as a design proceeds. Finally, design activity occurs within two contexts: the context within which the designer operates and the context produced by the developing design itself. The designer’s perception of what the context is affects the implication of the context on the design. The context shifts as the designer’s perceptions change. Design activity can be now characterized as a goal-oriented, con- strained, decision-making, exploration, and learning activity that operates within a con- text that depends on the designer’s percep- tion of the context.

Design Prototypes: A Knowledge Representation Schema for Design

Pattern recognition

Many sensor networks have been deployed to monitor Earth’s environment, and more will follow in the future. Environmental sensors have improved continuously by becoming smaller, cheaper, and more intelligent. Due to the large number of sensor manufacturers and differing accompanying protocols, integrating diverse sensors into observation systems is not straightforward. A coherent infrastructure is needed to treat sensors in an interoperable, platform-independent and uniform way. The concept of the Sensor Web reflects such a kind of infrastructure for sharing, finding, and accessing sensors and their data across different applications. It hides the heterogeneous sensor hardware and communication protocols from the applications built on top of it. The Sensor Web Enablement initiative of the Open Geospatial Consortium standardizes web service interfaces and data encodings which can be used as building blocks for a Sensor Web. This article illustrates and analyzes the recent developments of the new generation of the Sensor Web Enablement specification framework. Further, we relate the Sensor Web to other emerging concepts such as the Web of Things and point out challenges and resulting future work topics for research on Sensor Web Enablement.

New Generation Sensor Web Enablement

Maintenance: Changing role in life cycle management

- This paper describes the approach to building OntoSensor: a prototype sensor knowledge repository compatible with evolving Semantic Web infrastructure. OntoSensor includes definitions of concepts and properties adopted in part from SensorML, extensions to IEEE SUMO and references to ISO 19115. Simple queries have been developed and tested using Protege 2000 and Prolog. Although OntoSensor is in the early development stage, it presents a practical approach to building a sensor knowledge repository. It is proposed that OntoSensor may serve as a component in comprehensive applications that include more advanced inference mechanisms. Such comprehensive applications will be used for synergistic fusion of heterogeneous data in a network-centric environment. Keywords: Data Fusion, Semantic Web, Sensor Ontology 1.0 Introduction The assessment of situations and strategies in complex environments requires the fusion of information from heterogeneous data sources including sensors [1]. Synergistic fusion of data from multiple sensors ranging in complexity from simple acoustic sensors to sophisticated imaging equipment such as Forward Looking Infrared (FLIR) sensors will lead to the extraction of knowledge that cannot be perceived or inferred using individual sensors alone. For example, information from diverse sources such as sensors on an Unmanned Aerial Vehicle (UAV), reconnaissance reports and satellite imagery can be integrated to obtain high-level knowledge of objects in an area under surveillance including their spatial and temporal interrelationships; and to generate predictions of their intentions, positions and alignments in future states.

Building a Sensor Ontology: A Practical Approach Leveraging ISO and OGC Models.

This paper presents a practical approach to developing comprehensive sensor ontologies based upon deep knowledge models rather than capturing only superficial sensor attributes. It is proposed that the representation and utilization of deep sensor ontologies would enable a variety of sensor information system applications including sensor parts compatibility determination, dynamic sensor selection and tasking, and reasoning about systems of sensors in which data must be fused and queried from a variety of sensor types within a myriad of environments.

Sensor ontologies: from shallow to deep models

Existing image-based Captchas (completely automated public tests to tell computers and humans apart) utilize huge, public image databases to limit access to Websites via image matching attacks. In this paper, a different image based Captcha was developed and prototyped to address the mislabeling and other shortcomings of traditional schemes, which utilize huge public image databases. A database was populated with n 3D models. Then, random rotations, distortions, translations, lighting effects and warping were applied resulting in 2D images. Those 2D images were presented to users for identification to gain access to a Website. Users are prompted to identify the object with the label provided through a menu bar. This approach creates an infinite number of 2D images from a 3D model. Therefore, it appears to be impractical for an intruder to algorithmic ally identify the resulting 2D image without resorting to random guesses

2D Captchas from 3D Models

−The ability for software agents to discover, query, and task ubiquitous sensors requires machineinterpretable service descriptions, such as those proposed by the Semantic Web effort. Descriptions that support deep semantics will enable on-the-fly utilization of sensors for applications that might not have been anticipated on initial deployment. Semantic Web service discovery and dynamic composition requires formal semantic descriptions of inputs, outputs, preconditions, and effects of services. Universal Description Discovery and Integration (UDDI) provides a registry for publication and discovery of Web services, but it lacks the semantics needed for discovery and interoperation as envisioned by the Semantic Web community due to UDDI’s syntax-based search. This paper surveys representative approaches for incorporating semantic capabilities within the existing UDDI infrastructure and then proposes an architecture for sensor services within an ontology-based networkcentric environment.

Survey of Semantic Extensions to UDDI: Implications for Sensor Services.

This paper presents the design and test of a simple active near-infrared sparse detector imaging sensor. The prototype of the sensor is novel in that it can capture remarkable silhouettes or profiles of a wide-variety of moving objects, including humans, animals, and vehicles using a sparse detector array comprised of only sixteen sensing elements deployed in a vertical configuration. The prototype sensor was built to collect silhouettes for a variety of objects and to evaluate several algorithms for classifying the data obtained from the sensor into two classes: human versus non-human. Initial tests show that the classification of individually sensed objects into two classes can be achieved with accuracy greater than ninety-nine percent (99%) with a subset of the sixteen detectors using a representative dataset consisting of 512 signatures. The prototype also includes a Webservice interface such that the sensor can be tasked in a network-centric environment. The sensor appears to be a low-cost alternative to traditional, high-resolution focal plane array imaging sensors for some applications. After a power optimization study, appropriate packaging, and testing with more extensive datasets, the sensor may be a good candidate for deployment in vast geographic regions for a myriad of intelligent electronic fence and persistent surveillance applications, including perimeter security scenarios.

/pdf/sparse-detector-imaging-sensor-with-two-class-silhouette-2dhsmh5bdt.pdf

David J. Russomanno

Papers

Building a Sensor Ontology: A Practical Approach Leveraging ISO and OGC Models.

Sensor ontologies: from shallow to deep models

2D Captchas from 3D Models

Survey of Semantic Extensions to UDDI: Implications for Sensor Services.

Sparse Detector Imaging Sensor with Two-Class Silhouette Classification.