Machine Learning: New Ideas and Tools in Environmental Science and Engineering

Environmental Science & Technology, Journal Year: 2021, Volume and Issue: unknown

Published: Aug. 17, 2021

The rapid increase in both the quantity and complexity of data that are being generated daily field environmental science engineering (ESE) demands accompanied advancement analytics. Advanced analysis approaches, such as machine learning (ML), have become indispensable tools for revealing hidden patterns or deducing correlations which conventional analytical methods face limitations challenges. However, ML concepts practices not been widely utilized by researchers ESE. This feature explores potential to revolutionize modeling ESE field, covers essential knowledge needed applications. First, we use five examples illustrate how addresses complex problems. We then summarize four major types applications ESE: making predictions; extracting importance; detecting anomalies; discovering new materials chemicals. Next, introduce required current shortcomings ESE, with a focus on three important but often overlooked components when applying ML: correct model development, proper interpretation, sound applicability analysis. Finally, discuss challenges future opportunities application highlight this field.

Language: Английский

---

Machine Learning in IoT Security: Current Solutions and Future Challenges

IEEE Communications Surveys & Tutorials, Journal Year: 2020, Volume and Issue: 22(3), P. 1686 - 1721

Published: Jan. 1, 2020

The future Internet of Things (IoT) will have a deep economical, commercial and social impact on our lives. participating nodes in IoT networks are usually resource-constrained, which makes them luring targets for cyber attacks. In this regard, extensive efforts been made to address the security privacy issues primarily through traditional cryptographic approaches. However, unique characteristics render existing solutions insufficient encompass entire spectrum networks. Machine Learning (ML) Deep (DL) techniques, able provide embedded intelligence devices networks, can be leveraged cope with different problems. paper, we systematically review requirements, attack vectors, current We then shed light gaps these that call ML DL Finally, discuss detail addressing problems also several research directions ML- DL-based security.

Language: Английский

---

Modeling and forecasting building energy consumption: A review of data-driven techniques

Sustainable Cities and Society, Journal Year: 2019, Volume and Issue: 48, P. 101533 - 101533

Published: April 14, 2019

Language: Английский

---

Forecasting: theory and practice

International Journal of Forecasting, Journal Year: 2022, Volume and Issue: 38(3), P. 705 - 871

Published: Jan. 20, 2022

Forecasting has always been at the forefront of decision making and planning. The uncertainty that surrounds future is both exciting challenging, with individuals organisations seeking to minimise risks maximise utilities. large number forecasting applications calls for a diverse set methods tackle real-life challenges. This article provides non-systematic review theory practice forecasting. We provide an overview wide range theoretical, state-of-the-art models, methods, principles, approaches prepare, produce, organise, evaluate forecasts. then demonstrate how such theoretical concepts are applied in variety contexts. do not claim this exhaustive list applications. However, we wish our encyclopedic presentation will offer point reference rich work undertaken over last decades, some key insights practice. Given its nature, intended mode reading non-linear. cross-references allow readers navigate through various topics. complement covered by lists free or open-source software implementations publicly-available databases.

Language: Английский

---

Age of Information: A New Concept, Metric, and Tool

Foundations and Trends® in Networking, Journal Year: 2017, Volume and Issue: 12(3), P. 162 - 259

Published: Jan. 1, 2017

Age of information (AoI) was introduced in the early 2010s as a notion to characterize freshness knowledge system has about process observed remotely.AoI shown be fundamentally novel metric timeliness, significantly different, existing ones such delay and latency.The importance tool is paramount, especially contexts other than transport information, since communication takes place also control, or compute, infer, not just reproduce messages source.This volume comes present discuss first body works on AoI future directions that could yield more challenging interesting research.

Language: Английский

---

Interpreting Interpretability: Understanding Data Scientists' Use of Interpretability Tools for Machine Learning

Published: April 21, 2020

Machine learning (ML) models are now routinely deployed in domains ranging from criminal justice to healthcare. With this newfound ubiquity, ML has moved beyond academia and grown into an engineering discipline. To that end, interpretability tools have been designed help data scientists machine practitioners better understand how work. However, there little evaluation of the extent which these achieve goal. We study scientists' use two existing tools, InterpretML implementation GAMs SHAP Python package. conduct a contextual inquiry (N=11) survey (N=197) observe they uncover common issues arise when building evaluating models. Our results indicate over-trust misuse tools. Furthermore, few our participants were able accurately describe visualizations output by highlight qualitative themes for mental conclude with implications researchers tool designers, contextualize findings social science literature.

Language: Английский

---

A Survey on Security Threats and Defensive Techniques of Machine Learning: A Data Driven View

IEEE Access, Journal Year: 2018, Volume and Issue: 6, P. 12103 - 12117

Published: Jan. 1, 2018

Machine learning is one of the most prevailing techniques in computer science, and it has been widely applied image processing, natural language pattern recognition, cybersecurity, other fields. Regardless successful applications machine algorithms many scenarios, e.g., facial malware detection, automatic driving, intrusion these corresponding training data are vulnerable to a variety security threats, inducing significant performance decrease. Hence, vital call for further attention regarding threats defensive learning, which motivates comprehensive survey this paper. Until now, researchers from academia industry have found out against algorithms, including naive Bayes, logistic regression, decision tree, support vector (SVM), principle component analysis, clustering, deep neural networks. Thus, we revisit existing give systematic on them two aspects, phase testing/inferring phase. After that, categorize current into four groups: assessment mechanisms, countermeasures phase, those testing or inferring security, privacy. Finally, provide five notable trends research worth doing in-depth studies future.

Language: Английский

---

Social media big data analytics: A survey

Computers in Human Behavior, Journal Year: 2018, Volume and Issue: 101, P. 417 - 428

Published: Aug. 22, 2018

Language: Английский

---

Challenges and Future Directions of Big Data and Artificial Intelligence in Education

Frontiers in Psychology, Journal Year: 2020, Volume and Issue: 11

Published: Oct. 19, 2020

We discuss the new challenges and directions facing use of big data artificial intelligence (AI) in education research, policy-making, industry. In recent years, applications AI have made significant headways. This highlights a novel trend leading-edge educational research. The convenience embeddedness collection within technologies, paired with computational techniques analyses reality. are moving beyond proof-of-concept demonstrations techniques, beginning to see substantial adoption many areas education. key research trends domains associated assessment, individualized learning, precision Model-driven analytics approaches will grow quickly guide development, interpretation, validation algorithms. However, conclusions from should, course, be applied caution. At policy level, government should devoted supporting lifelong offering teacher programs, protecting personal data. With regard industry, reciprocal mutually beneficial relationships developed order enhance academia-industry collaboration. Furthermore, it is important make sure that technologies guided by relevant theoretical frameworks empirically tested. Lastly, this paper we advocate an in-depth dialogue between supporters “cold” technology “warm” humanity so can lead greater understanding among teachers students about how technology, specifically, explosion revolution bring opportunities (and challenges) best leveraged for pedagogical practices learning.

Language: Английский

---

ModAugNet: A new forecasting framework for stock market index value with an overfitting prevention LSTM module and a prediction LSTM module

Expert Systems with Applications, Journal Year: 2018, Volume and Issue: 113, P. 457 - 480

Published: July 9, 2018

Language: Английский

---