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 Practical Machine Learning Tools and Techniques

物件導向軟體之架構(Object-Oriented Software Construction)探討

In the real world, a realistic setting for computer vision or multimedia recognition problems is that we have some classes containing lots of training data and many classes contain a small amount of training data. Therefore, how to use frequent classes to help learning rare classes for which it is harder to collect the training data is an open question. Learning with Shared Information is an emerging topic in machine learning, computer vision and multimedia analysis. There are different level of components that can be shared during concept modeling and machine learning stages, such as sharing generic object parts, sharing attributes, sharing transformations, sharing regularization parameters and sharing training examples, etc. Regarding the specific methods, multi-task learning, transfer learning and deep learning can be seen as using different strategies to share information. These learning with shared information methods are very effective in solving real-world large-scale problems. This special issue aims at gathering the recent advances in learning with shared information methods and their applications in computer vision and multimedia analysis. Both state-of-the-art works, as well as literature reviews, are welcome for submission. Papers addressing interesting real-world computer vision and multimedia applications are especially encouraged. Topics of interest include, but are not limited to:  • Multi-task learning or transfer learning for large-scale computer vision and multimedia analysis • Deep learning for large-scale computer vision and multimedia analysis • Multi-modal approach for large-scale computer vision and multimedia analysis • Different sharing strategies, e.g., sharing generic object parts, sharing attributes, sharing transformations, sharing regularization parameters and sharing training examples, • Real-world computer vision and multimedia applications based on learning with shared information, e.g., event detection, object recognition, object detection, action recognition, human head pose estimation, object tracking, location-based services, semantic indexing. • New datasets and metrics to evaluate the benefit of the proposed sharing ability for the specific computer vision or multimedia problem. • Survey papers regarding the topic of learning with shared information.  Authors who are unsure whether their planned submission is in scope may contact the guest editors prior to the submission deadline with an abstract, in order to receive feedback.

IEEE transactions on pattern analysis and machine intelligence

The Object Management Group's (OMG) Model Driven Architecture (MDA) strategy envisages a world where models play a more direct role in software production, being amenable to manipulation and transformation by machine. Model Driven Engineering (MDE) is wider in scope than MDA. MDE combines process and analysis with architecture. This article sets out a framework for model driven engineering, which can be used as a point of reference for activity in this area. It proposes an organisation of the modelling 'space' and how to locate models in that space. It discusses different kinds of mappings between models. It explains why process and architecture are tightly connected. It discusses the importance and nature of tools. It identifies the need for defining families of languages and transformations, and for developing techniques for generating/configuring tools from such definitions. It concludes with a call to align metamodelling with formal language engineering techniques.

Model Driven Engineering

A three-valve pulsatile tubular pump was used in 10 pigs weighing 9.9+/-0.3 kg, connected to a neonatal hemofiltration (HF) circuit to analyze the parameters to determine the ultrafiltration effectiveness for venovenous continuous renal replacement therapy. Forty 30-min periods were studied to analyze the influence of the catheter diameter, purification technique (HF or hemodiafiltration), pump pulsation frequency, percentage time in diastole, percentage time in diastole in which the post pump valve was closed, and percentage time in systole in which the post filter valve was closed, on ultrafiltrate volume and blood flow through the filter. A higher ultrafiltrate flow was achieved with a frequency of 60 bpm than with 30 bpm, a diastolic time of 65% of the total cycle versus 50%, and a blood flow of greater than 20 mL/min versus less than 20 mL/min. The ultrafiltrate flow increases in relation to the blood flow, the frequency of the pump, and the diastolic time.

Continuous Venovenous Renal Replacement Therapy With a Pulsatile Tubular Blood Pump: Analysis of Efficacy Parameters

Back in 2008 GMV released the first version of its magicGNSS suite, an online set of GNSS tools which have been tailored throughout the years to cope with the needs of a wide variety of GNSS users (over 1000 registered users by 2013), providing added-value services including multi-GNSS Precise Orbit and Clock Determination, High Accuracy Positioning, Integrity and local augmentation developments among others.

The precise orbit and clock products for the different satellite constellations are generated in an Orbit Determination and Time Synchronisation (ODTS) process. This process receives as input dual-frequency code and phase measurements from a network of reference stations and produces as output precise satellite orbits and clocks, together with additional estimated parameters such as station clock biases, tropospheric delays and phase ambiguities. All GNSS satellites are processed together to ensure a consistent solution is provided. Due to GLONASS satellites transmitting individual frequencies (FDMA as opposed to CDMA used by GPS and Galileo), an inter-frequency bias must be estimated when processing GLONASS data together with the inter-system bias for Galileo. The quality of the multi-GNSS orbit and clock products generated by magicGNSS have been assessed by means of their comparison with the products generated by different Analysis Centers (TUM, CODE, CNES and GFZ) in the frame of IGS’ Multi-GNSS Experiment (MGEX) project with centimetric differences.

The Precise Point Positioning (PPP) capability of magicGNSS suite provides centimetric positioing error, and sub-nanosecond error in timing, using dual-frequency observations in RINEX format from a stand-alone user receiver and precise GNSS orbit and clock products calculated beforehand (products from the International GNSS Service (IGS), multi-GNSS products generated by GMV or orbits and clocks generated by the user). The user receiver can be fixed to the ground (static PPP) or can be a moving receiver (dynamic or kinematic PPP). One of the main advantages of PPP with respect to other precise-positioning approaches like Real Time Kinematics (RTK) in that no ‘base station’ or surrounding stations are needed. Another advantage is that since the GNSS orbit and clock products are by nature global, and therefore, the PPP solutions are also global.

Together with the magicGNSS online suite, GMV has been developing over the last two years an infrastructure for the generation of precise GPS, GLONASS and Galileo orbits and clocks in real-time. This infrastructure acquires via NTRIP data streams from around eighty IGS tracking stations distributed worldwide, and produces orbit updates every fifteen minutes and clock updates every second from a combined multi-GNSS solution that can be then used consistently for PPP applications. These real-time satellite orbits and clocks are available as a real-time ephemeris correction stream in standard RTCM format and can be used for real-time PPP using GMV’s magicGNSS PPP demonstrator or any real-time PPP client which processes RTCM ephemeris corrections. This real-time High-Accuracy (HA) service is comparable to the IGS service; the clock RMS is around 0.3 nanoseconds compared to the IGS rapid product (75ps RMS) and the prediction orbit is around 6 cm of the IGS rapid orbit product (~2.5 cm).

GMV’s magicGNSS real-time PPP demonstrator was developed in parallel to the real-time HA multi-GNSS products generation platform to evaluate real-time PPP performance in realistic scenarios (kinematic and static). It provides an end to end process including redundancy, communications, robustness and reliability and the ability to run on portable devices. The system has already been evaluated under several field scenarios representing many situations that potential users could address in real operations for many different applications. These include static, kinematic and combined use cases. In the carried out tests, different visibility conditions have been evaluated (open sky or different types of obstacles such as trees or walls), as well as the robustness of the solution against communication losses of different durations. The real-time PPP solutions are validated against RTK and/or post-processed PPP.

As an example of the aforementioned tests, a receiver was mounted onto the roof of a car to test the real-time PPP solution in kinematic mode. This test showed that during the collection of data (a 15 km route around a base station) the solution degraded when a lower amount of satellites were in view due to obstacles such as trees and buildings. This degradation can result in a re-estimation of the ambiguities which can take up to 20 minutes to achieve. It was noticed that the inclusion of multiple GNSS observations during this testing was an advantage as more satellites were in view with respect to the GPS-only case, limiting the instances of a re-estimation of the ambiguities being needed.

Currently intensive research is being carried out at GMV for developing strategies to reduce convergence times, which are normally around 30 minutes for decimetric accuracy. One such strategy that is used is the ‘quick start’ feature which involves starting at a point with calibrated (few centimetres) a-priori coordinates. Use of this feature can lead to convergence within a few seconds which is crucial for certain PPP users.

Throughout this paper, magicGNSS` real-time orbits and clocks server and the real-time PPP demonstrator developments undertaken are described, together with both the server (i.e. orbit and clock) performances achieved with respect to IGS and the resulting positioning performances. We also discuss the major challenges faced in the development process, and some under research ways to improve the overall service.

magicGNSS’ Real-Time POD and PPP Multi-GNSS Service

Semi-Supervised Learning (SSL) is a data mining technique which comes between supervised and unsupervised techniques, and is useful when a small number of instances in a dataset are labelled but a lot of unlabelled data is also available. This is the case with user reviews in application stores such as the Apple App Store or Google Play, where a vast amount of reviews are available but classifying them into categories such as bug related review or feature request is expensive or at least labor intensive. SSL techniques are well-suited to this problem as classifying reviews not only takes time and effort, but may also be unnecessary. In this work, we analyse SSL techniques to show their viability and their capabilities in a dataset of reviews collected from the App Store for both transductive (predicting existing instance labels during training) and inductive (predicting labels on unseen future data) performance.

Preliminary Study on Applying Semi-Supervised Learning to App Store Analysis

The emerging consensus on the boundaries and main elements of the Software Engineering (SE) discipline represents an opportunity for the engineering of shared conceptualizations that may serve both to design automated tools and tasks that help in diverse phases and aspects of the software process, and also to annotate learning–oriented resources Formal ontologies provide an appropriate logics– based framework for such conceptual models This paper describes the main ontological commitments that underlie the Onto-SWEBOK project, focusing on how SE artifacts and activities can be represented The paper also discusses how semantic annotations can be provided for learning resources oriented to the initial and continuing education on the discipline, which enables the reuse of such resources in diverse learning designs

/pdf/ontologies-of-software-artifacts-and-activities-resource-h0q4mtm0ow.pdf

Ontologies of software artifacts and activities: Resource annotation and application to learning technologies

This work proposes a biometric identification system that works together with a palm vein reader sensor and a hand-clenching support, designed to perform the capture the back of the hand. Several processing steps were performed: extraction of the region of interest, binarization, dilation, noise filtering, skeletonization, as well as extraction and verification of patterns based on the measurment of coincidence of vertical and horizontal displacements of skeletonized and dilated images. The proposed method achieved the following results: processing time post capture of 1.8 seconds, FRR of 0.47% and FAR of 0,00%, with a referential database of 50 people from a total of 1500 random captures.

/pdf/a-biometric-method-based-on-the-matching-of-dilated-and-sq3dpjwhjc.pdf

Daniel Rodriguez

Papers

Continuous Venovenous Renal Replacement Therapy With a Pulsatile Tubular Blood Pump: Analysis of Efficacy Parameters

magicGNSS’ Real-Time POD and PPP Multi-GNSS Service

Preliminary Study on Applying Semi-Supervised Learning to App Store Analysis

Ontologies of software artifacts and activities: Resource annotation and application to learning technologies

A biometric method based on the matching of dilated and skeletonized IR images of the veins map of the dorsum of the hand