China
Africa
Explore chapters and articles related to this topic
Machine Learning
Published in Anil Kumar, Priyadarshi Upadhyay, A. Senthil Kumar, Fuzzy Machine Learning Algorithms for Remote Sensing Image Classification, 2020
Anil Kumar, Priyadarshi Upadhyay, A. Senthil Kumar
The pattern recognition approach provides a solution for given inputs and performs matching of the inputs while considering their statistical variation. It is unique in comparison to pattern matching approaches, as it finds exact matches in inputs with past existing patterns. Regular expression matching is one of the examples of a pattern matching algorithm, which identifies patterns in textual data. This data is utilized in search capabilities of large text editors and word processors.
Decisions and Decision-Making
Published in Joseph Eli Kasser, Systems Thinker’s Toolbox, 2018
Since the decision-making process overlaps the problem-solving process up to the point where the decision is made, the decision-making process provides a different perspective on the grouping of the activities performed in problem-solving and decision-making. Since different perspectives provide different insights, perceive the process from the perspective of the following four key elements10 in making decisions (Russo and Schoemaker 1989: p. 2): Framing the problem: considering what must be decided. This is the most critical part of the process because if the problem is framed incorrectly, then the wrong problem will be considered, the wrong questions will be asked, and the wrong solution will be realized. A Problem Formulation Template for framing the problem is presented and discussed in Section 14.3.Gathering intelligence: determining what factors are pertinent in the situation and what factors can be safely ignored. The pertinent factors relate to determining the solution options and the selection criteria for making the decision. The Five Whys (Section 7.5) is one way of determining what intelligence needs to be gathered.Coming to conclusions: a sound systemic and systematic approach considering all the parameters generally results in a better decision. This is the application of Critical Thinking (Chapter 3).Learning (or failing to learn) from previous decisions: the application of the principles of Lessons Learned (Section 9.3) pattern matching, feedback, and improvement. You need to compare the expected outcomes of the decision with the actual outcomes and learn from the differences and understand the reasons for the differences. Pattern matching allows you to compare the current situation in which you are making the decision with other similar situations where the patterns in the data match and facilitate the decision. Feedback is the principle of closing the loop and learning from the effects of the decision (the good and the bad results) and not making the same mistake in the future. Not making the same mistake in the future leads to improvements in your decision-making which will result in better decisions.
A Big Data Analytics-driven Lean Six Sigma framework for enhanced green performance: a case study of chemical company
Published in Production Planning & Control, 2023
Amine Belhadi, Sachin S. Kamble, Angappa Gunasekaran, Karim Zkik, Dileep Kumar M., Fatima Ezahra Touriki
To ensure that the prototyping was performed rigorously, four criteria around repeatability, reliability, and validity were applied as follows:Repeatability: The selection of cases included a variety of tiers, sizes, and locations. A cross-comparison of experts' inputs avoids accidental interference and finds the similarities and differences among all experts' inputs.Reliability: multiple sources of inputs (experts' opinion and literature review) was used. Triangulation of the findings was performed using internal and external documents of companies, formal debriefing, status reports, and surveys. A predefined protocol of questions was followed for all interviews. More experts were consulted until discussions were not able to provide additional insights or meaningful information.Validity: a wide variety was considered within the experts' panel, including different locations and sizes. Several rounds of data checking and triangulation with literature review findings were performed. Coding was used to enhance pattern matching.
Optimal multi-threshold quantization scheme for bioinformatics inspired cooperative spectrum sensing in cognitive radio networks
Published in International Journal of Electronics, 2018
In quantization based CSS, the CR users quantize the received energy into quantization zones based on quantization thresholds and send multiple bits instead of one bit to the FC. The performance of quantization based CSS is dependent upon the quantization parameters. Energy zones reflect the PU activity and if the quantization parameters are not optimal the PU activity will not be reflected accurately. If the low received energies are represented through higher quantization zones it results in higher probability of false alarm and representing high received energies with low quantization zones results in higher probability of misdetection. The selection of the quantization thresholds should satisfy certain constraints of system performance. Local decisions of CR users are combined at the fusion center (FC) to take a global decision. Local decisions of some CR users may be affected by fading and thus are different from the rest of the CR users. Including the local decisions of such CR users in global decision, combination affects the sensing performance. The scenario where the CR users have unreliable reports and the FC needs to identify the unreliable CR users to exclude their reports from the final decision combination, the methods of bioinformatics can be used. String matching is used extensively for biological sequences alignment in bioinformatics. The string matching is pattern matching, and the techniques developed for biological sequence alignment in bioinformatics can be used in CRN. Both the processes are similar as both are concerned with detecting anomalies in patterns of information.
Design of a speech-enabled 3D marine compass simulation system
Published in Ships and Offshore Structures, 2018
Bin Fu, Hongxiang Ren, Jingjing Liu, Xiaoxi Zhang
Speech recognition is essentially a type of pattern recognition system that consists primarily of three basic elements: feature extraction, pattern matching and a reference model library. Its core is to construct a mapping relation between the speech feature vector sequence and the reference character sequence of the model. The most common types of speech recognition systems are speaker-dependent speech recognition, speaker-independent speech recognition, isolated-word speech recognition and continuous speech recognition. Among these, isolated-word recognition uses dynamic time warping to solve the problem of matching the varying lengths between the feature vector sequences of a reference template and an input speech feature vector sequence (Myers et al. 1980). The hidden Markov model is used to establish an acoustic model for both isolated word and continuous speech recognition. This model has many years of application history and is a core algorithm used in speech recognition (Zarrouk and Ayed 2014). However, deep learning has become another research hotspot because of growing demand for large-vocabulary continuous speech recognition systems. Additionally, deep neural networks and convolutional neural networks are representative of the research methods for large-vocabulary continuous speech recognition, and they have achieved good results in various applications (Sainath et al. 2013; Maas et al. 2017). The basic framework for a speech recognition system based on pattern matching is shown in Figure 7. Below, we provide an analysis of the signal preprocessing, feature extraction and speech recognition decoding operations.