A transferable active-learning strategy for reactive molecular force fields

SALAD: An Exploration of Split Active Learning based Unsupervised Network Data Stream Anomaly Detection using Autoencoders

10.36227/techrxiv.14896773 ◽

2021 ◽

Author(s):

Christopher Nixon ◽

Mohamed Sedky ◽

Mohamed Hassan

Keyword(s):

Machine Learning ◽

Active Learning ◽

Intrusion Detection ◽

Anomaly Detection ◽

Data Streams ◽

Data Stream ◽

Learning Strategy ◽

Network Data ◽

Anomaly Detector ◽

Active Learning Strategy

<div>Machine learning based intrusion detection systems monitor network data streams for cyber attacks. Challenges in this space include detection of unknown attacks, adaptation to changes in the data stream such as changes in underlying behaviour, the human cost of labeling data to retrain the machine learning model and the processing and memory constraints of a real-time data stream. Failure to manage the aforementioned factors could result in missed attacks, degraded detection performance, unnecessary expense or delayed detection times. This research evaluated autoencoders, a type of feed-forward neural network, as online anomaly detectors for network data streams. The autoencoder method was combined with an active learning strategy to further reduce labeling cost and speed up training and adaptation times, resulting in a proposed Split Active Learning Anomaly Detector (SALAD) method. The proposed method was evaluated with the NSL-KDD, KDD Cup 1999, and UNSW-NB15 data sets, using the scikit-multiflow framework. Results demonstrated that a novel Adaptive Anomaly Threshold method, combined with a split active learning strategy offered superior anomaly detection performance with a labeling budget of just 20%, significantly reducing the required human expertise to annotate the network data. Processing times of the autoencoder anomaly detector method were demonstrated to be significantly lower than traditional online learning methods, allowing for greatly improved responsiveness to attacks occurring in real time. Future research areas are applying unsupervised threshold methods, multi-label classification, sample annotation, and hybrid intrusion detection.</div>

Download Full-text

A Transferable Active-Learning Strategy for Reactive Molecular Force Fields

10.26434/chemrxiv.13856123 ◽

2021 ◽

Author(s):

Tom Young ◽

Tristan Johnston-Wood ◽

Volker Deringer ◽

Fernanda Duarte

Keyword(s):

Active Learning ◽

Large Scale ◽

Learning Strategy ◽

Gaussian Approximation ◽

Force Fields ◽

Molecular System ◽

Chemical Reactivity ◽

Bulk Water ◽

Interatomic Potentials ◽

Molecular Systems

<p>Predictive simulations of dynamic processes in molecular systems require fast, accurate and reactive interatomic potentials. Machine learning offers a promising approach to construct force-field models for large-scale molecular simulation by fitting to high-level quantum-mechanical data. However, machine-learned force fields generally require considerable human intervention and data volume. Here we show that, by leveraging hierarchical and active learning, accurate Gaussian Approximation Potential (GAP) models for diverse chemical systems can be developed in an autonomous way, requiring only hundreds to a few thousand energy and gradient evaluations on the reference potential-energy surface. Our approach relies on a decomposition of the condensed-phase molecular system into intra- and inter-molecular terms, and on the definition of a prospective error metric to quantify accuracy. We demonstrate applications to a range of molecular systems: from bulk water, organic solvents, and a solvated ion onwards to the description of chemical reactivity, including, a bifurcating Diels–Alder reaction in the gas phase and non-equilibrium dynamics (S<sub>N</sub>2 reaction) in explicit solvent. The method provides a route to routinely generating machine-learned force fields for complex and/or reactive molecular systems. </p>

Download Full-text

A Transferable Active-Learning Strategy for Reactive Molecular Force Fields

10.26434/chemrxiv.13856123.v1 ◽

2021 ◽

Author(s):

Tom Young ◽

Tristan Johnston-Wood ◽

Volker Deringer ◽

Fernanda Duarte

Keyword(s):

Active Learning ◽

Large Scale ◽

Learning Strategy ◽

Gaussian Approximation ◽

Force Fields ◽

Molecular System ◽

Chemical Reactivity ◽

Bulk Water ◽

Interatomic Potentials ◽

Molecular Systems

<p>Predictive simulations of dynamic processes in molecular systems require fast, accurate and reactive interatomic potentials. Machine learning offers a promising approach to construct force-field models for large-scale molecular simulation by fitting to high-level quantum-mechanical data. However, machine-learned force fields generally require considerable human intervention and data volume. Here we show that, by leveraging hierarchical and active learning, accurate Gaussian Approximation Potential (GAP) models for diverse chemical systems can be developed in an autonomous way, requiring only hundreds to a few thousand energy and gradient evaluations on the reference potential-energy surface. Our approach relies on a decomposition of the condensed-phase molecular system into intra- and inter-molecular terms, and on the definition of a prospective error metric to quantify accuracy. We demonstrate applications to a range of molecular systems: from bulk water, organic solvents, and a solvated ion onwards to the description of chemical reactivity, including, a bifurcating Diels–Alder reaction in the gas phase and non-equilibrium dynamics (S<sub>N</sub>2 reaction) in explicit solvent. The method provides a route to routinely generating machine-learned force fields for complex and/or reactive molecular systems. </p>

Download Full-text

A Transferable Active-Learning Strategy for Reactive Molecular Force Fields

10.26434/chemrxiv.13856123.v2 ◽

2021 ◽

Author(s):

Tom Young ◽

Tristan Johnston-Wood ◽

Volker Deringer ◽

Fernanda Duarte

Keyword(s):

Active Learning ◽

Large Scale ◽

Learning Strategy ◽

Gaussian Approximation ◽

Force Fields ◽

Molecular System ◽

Chemical Reactivity ◽

Bulk Water ◽

Interatomic Potentials ◽

Molecular Systems

Predictive simulations of dynamic processes in molecular systems require fast, accurate and reactive interatomic potentials. Machine learning offers a promising approach to construct force-field models for large-scale molecular simulation by fitting to high-level quantum-mechanical data. However, machine-learned force fields generally require considerable human intervention and data volume. Here we show that, by leveraging hierarchical and active learning, accurate Gaussian Approximation Potential (GAP) models for diverse chemical systems can be developed in an autonomous way, requiring only hundreds to a few thousand energy and gradient evaluations on the reference potential-energy surface. Our approach relies on a decomposition of the condensed-phase molecular system into intra- and inter-molecular terms, and on the definition of a prospective error metric to quantify accuracy. We demonstrate applications to a range of molecular systems: from bulk water, organic solvents, and a solvated ion onwards to the description of chemical reactivity, including, a bifurcating Diels–Alder reaction in the gas phase and non-equilibrium dynamics (SN2 reaction) in explicit solvent. The method provides a route to routinely generating machine-learned force fields for complex and/or reactive molecular systems.

Download Full-text

SALAD: An Exploration of Split Active Learning based Unsupervised Network Data Stream Anomaly Detection using Autoencoders

10.36227/techrxiv.14896773.v1 ◽

2021 ◽

Author(s):

Christopher Nixon ◽

Mohamed Sedky ◽

Mohamed Hassan

Keyword(s):

Machine Learning ◽

Active Learning ◽

Intrusion Detection ◽

Anomaly Detection ◽

Data Streams ◽

Data Stream ◽

Learning Strategy ◽

Network Data ◽

Anomaly Detector ◽

Active Learning Strategy

<div>Machine learning based intrusion detection systems monitor network data streams for cyber attacks. Challenges in this space include detection of unknown attacks, adaptation to changes in the data stream such as changes in underlying behaviour, the human cost of labeling data to retrain the machine learning model and the processing and memory constraints of a real-time data stream. Failure to manage the aforementioned factors could result in missed attacks, degraded detection performance, unnecessary expense or delayed detection times. This research evaluated autoencoders, a type of feed-forward neural network, as online anomaly detectors for network data streams. The autoencoder method was combined with an active learning strategy to further reduce labeling cost and speed up training and adaptation times, resulting in a proposed Split Active Learning Anomaly Detector (SALAD) method. The proposed method was evaluated with the NSL-KDD, KDD Cup 1999, and UNSW-NB15 data sets, using the scikit-multiflow framework. Results demonstrated that a novel Adaptive Anomaly Threshold method, combined with a split active learning strategy offered superior anomaly detection performance with a labeling budget of just 20%, significantly reducing the required human expertise to annotate the network data. Processing times of the autoencoder anomaly detector method were demonstrated to be significantly lower than traditional online learning methods, allowing for greatly improved responsiveness to attacks occurring in real time. Future research areas are applying unsupervised threshold methods, multi-label classification, sample annotation, and hybrid intrusion detection.</div>

Download Full-text

Effectiveness of active learning strategy in improving the acoustic awareness skills and understanding what is heard by the basic stage students in Jordan

Educational Research and Reviews ◽

10.5897/err2016.2944 ◽

2016 ◽

Vol 11 (20) ◽

pp. 1896-1905

Author(s):

Al-Odwan Yaser

Keyword(s):

Active Learning ◽

Learning Strategy ◽

Basic Stage ◽

Active Learning Strategy

Download Full-text

Enhancing generic capabilities and metacognitive awareness of first-year nursing students using active learning strategy

BMC Nursing ◽

10.1186/s12912-021-00601-7 ◽

2021 ◽

Vol 20 (1) ◽

Author(s):

Carmen Wing Han Chan ◽

Fiona Wing Ki Tang ◽

Ka Ming Chow ◽

Cho Lee Wong

Keyword(s):

Nursing Education ◽

Nursing Students ◽

Active Learning ◽

Flipped Classroom ◽

Learning Strategy ◽

Interactive Learning ◽

Metacognitive Awareness ◽

University Education ◽

Strong Impact ◽

Active Learning Strategy

Abstract Background Developing students’ generic capabilities is a major goal of university education as it can help to equip students with life-long learning skills and promote holistic personal development. However, traditional didactic teaching has not been very successful in achieving this aim. Kember and Leung’s Teaching and Learning Model suggests an interactive learning environment has a strong impact on developing students’ generic capabilities. Metacognitive awareness is also known to be related to generic capability development. This study aimed to assess changes on the development of generic capabilities and metacognitive awareness after the introduction of active learning strategy among nursing students. Methods This study adopted a quasi-experimental single group, matched pre- and posttest design. It was conducted in a school of nursing at a university in Hong Kong. Active learning approaches included the flipped classroom (an emphasis on pre-reading) and enhanced lectures (the breaking down of a long lecture into several mini-lectures and supplemented by interactive learning activities) were introduced in a foundational nursing course. The Capabilities Subscale of the Student Engagement Questionnaire and the Metacognitive Awareness Inventory were administered to two hundred students at the start (T0) and at the end of the course (T1). A paired t-test was performed to examine the changes in general capabilities and metacognitive awareness between T0 and T1. Results A total of 139 paired pre- and post-study responses (69.5 %) were received. Significant improvements were observed in the critical thinking (p < 0.001), creative thinking (p = 0.03), problem-solving (p < 0.001) and communication skills (p = 0.04) with the implementation of active learning. Significant changes were also observed in knowledge of cognition (p < 0.001) and regulation of cognition (p < 0.001) in the metacognitive awareness scales. Conclusions Active learning is a novel and effective teaching approach that can be applied in the nursing education field. It has great potential to enhance students’ development of generic capabilities and metacognitive awareness.

Download Full-text