Driven Machine Learning Research Articles

Development of data driven machine learning models for the prediction and design of pyrimidine corrosion inhibitors

Pyrimidines have been shown as promising nontoxic corrosion inhibitors for carbon steel in acid media that can replace toxic chemicals currently in use. However, the discovery of this important corrosion inhibitor has mainly be conducted by expensive trial and error experimental approaches. However, recent studies indicates that the use of machine learning can help in the speedy discovery of novel corrosion inhibitor molecules with minimal cost. In the present work, machine learning algorithms were utilized to develop predictive models for fifty-four (54) pyrimidines derivatives whose experimentally determined inhibition efficiencies data as corrosion inhibitors for carbon steel in hydrochloric acid medium are available in the literature utilizing the partial least square regression (PLS) and the random forest (RF). Seven descriptors were selected by PLS and used to develop the linear model. The variable importance results using the PLS indicates that molecular mass, molecular volume, electrophilicity, electronegativity, energy of the lowest unoccupied molecular orbital, electron affinity, and the logarithm of the partition coefficient were the main factors that determine the inhibition efficiencies. RF was used to capture the nonlinear nature of the data and to accurately predict the inhibition efficiency. Rigorous internal and external validation were performed using the PLS and RF to further verify the robustness and predictive ability of the models. The random forest yielded the best results with the mean standard error (MSE) of 32.602 compared to the PLS with MSE of 64.641. Both models were subsequently used for the prediction of five (5) new pyrimidines with very high inhibition efficiency. The result of this work can provide reference information and theoretical guidance for designing and synthesizing new and effective pyrimidines corrosion inhibitors.

Assessing the potential of machine learning methods to study the removal of pharmaceuticals from wastewater using biochar or activated carbon

Pharmaceuticals in wastewater are rapidly becoming new emerging pollutants, affecting humans and the aquatic ecosystem, and can go undetected due to their microscopic nature. Adsorption proves to be a promising technology for the removal of pharmaceuticals from effluent wastewater owning to its low cost, flexibility, and renewability. Adsorbents are porous materials such as silica, clay, resins, and carbon-based materials (e.g., biochars, carbon nanotubes, and activated carbon) often used to remove pharmaceutical micropollutants during adsorption. Among them biochar is an emerging, cost-effective, and eco-friendly sorbent. Modeling methods such as linear correlativity and multilinear regressions, are often employed to explain the adsorption mechanism, however they show limited accuracy and applicability. On the contrary, data driven machine learning (ML) methods is a powerful tool that could be used to study the complex relationship between adsorption performances and biochar properties. This review provides an overview of recent advances in the use of machine learning (ML) methods to explore the field of pharmaceutical adsorption onto biochar. An introduction to different ML algorithms and their advantages and limitations is provided. Furthermore, the challenges and future prospects of ML applications to study the adsorption mechanism is outlined.

Neuroevolution Guided Hybrid Spiking Neural Network Training.

Neuromorphic computing algorithms based on Spiking Neural Networks (SNNs) are evolving to be a disruptive technology driving machine learning research. The overarching goal of this work is to develop a structured algorithmic framework for SNN training that optimizes unique SNN-specific properties like neuron spiking threshold using neuroevolution as a feedback strategy. We provide extensive results for this hybrid bio-inspired training strategy and show that such a feedback-based learning approach leads to explainable neuromorphic systems that adapt to the specific underlying application. Our analysis reveals 53.8, 28.8, and 28.2% latency improvement for the neuroevolution-based SNN training strategy on CIFAR-10, CIFAR-100, and ImageNet datasets, respectively in contrast to state-of-the-art conversion based approaches. The proposed algorithm can be easily extended to other application domains like image classification in presence of adversarial attacks where 43.2 and 27.9% latency improvements were observed on CIFAR-10 and CIFAR-100 datasets, respectively.

Predicting liner wear of ball mills using discrete element method and artificial neural network

Mill component wear is a critical issue in industrial grinding mills as it affects mill’s continuous operation and performance and causes a considerable cost to replace worn parts. Understanding wear profile and its effects on mill performance would provide useful insight for process optimisation. In this work, an approach based on the Discrete Element Method (DEM) and data driven machine learning model has been proposed to predict the liner wear profile of ball mills. DEM simulations based on a small scaled experimental setup were carried out to examine the relationship between liner wear energy and the input features such as the location of the point considered on the liner, the ratio of particle size to mesh element size and energy distribution over the wear profile. Then the machine learning model for wear parameter was developed. The comparison between the actual and predicted wear parameters for the test data showed a correlation coefficient of about 0.95 and an accuracy of 93% with the desired outcomes. The predicted liner wear profiles at different times for the mill were similar to the ones obtained by the experiment. The approach proposed was finally applied to a large industrial mill with a diameter of 4.267 m. It was shown that only five intermediate simulations were needed to determine a liner wear profile of the mill close to that from field data after 57,726 h of operation. The results obtained in this study indicate that combining neural networks with the DEM provides a promising technique in developing wear models for industrial applications.

The NMT Scalp EEG Dataset: An Open-Source Annotated Dataset of Healthy and Pathological EEG Recordings for Predictive Modeling

Electroencephalogram (EEG) is widely used for the diagnosis of neurological conditions like epilepsy, neurodegenerative illnesses and sleep related disorders. Proper interpretation of EEG recordings requires the expertise of trained neurologists, a resource which is scarce in the developing world. Neurologists spend a significant portion of their time sifting through EEG recordings looking for abnormalities. Most recordings turn out to be completely normal, owing to the low yield of EEG tests. To minimize such wastage of time and effort, automatic algorithms could be used to provide pre-diagnostic screening to separate normal from abnormal EEG. Data driven machine learning offers a way forward however, design and verification of modern machine learning algorithms require properly curated labeled datasets. To avoid bias, deep learning based methods must be trained on large datasets from diverse sources. This work presents a new open-source dataset, named the NMT Scalp EEG Dataset, consisting of 2,417 recordings from unique participants spanning almost 625 h. Each recording is labeled as normal or abnormal by a team of qualified neurologists. Demographic information such as gender and age of the patient are also included. Our dataset focuses on the South Asian population. Several existing state-of-the-art deep learning architectures developed for pre-diagnostic screening of EEG are implemented and evaluated on the NMT, and referenced against baseline performance on the well-known Temple University Hospital EEG Abnormal Corpus. Generalization of deep learning based architectures across the NMT and the reference datasets is also investigated. The NMT dataset is being released to increase the diversity of EEG datasets and to overcome the scarcity of accurately annotated publicly available datasets for EEG research.

MetaAnalyser - A Concept and Toolkit for Enablement of Digital Twin

Digital Twin (DT) has promising impact on the life cycle management of assets in manufacturing industry. The concept of DT has become possible with digitalisation and Artificial Intelligence (AI). Data driven Machine Learning (ML) capabilities, can enhance the performance of the DT. To replicate a dynamic system, the DT should continuously receive and process incoming data in real-time. However, every time that the system receives new incoming datasets, the challenges of ML such as data preparation, feature selection, model selection and performance evaluation, slow down the development process of DT. This paper proposes a MetaAnalyser platform that automates these steps for incoming datasets in real-time. The MetaAnalyser platform through automating data preparation, feature selection, model selection and performance evaluation, is expected to increase the level of agility in the development process of DT and the efficiency of the DT during its lifecycle. The MetaAnalyser platform is demonstrated in this paper by ranking the features that affect the arrival delays in trains and ranking regression models based on their performance on the dataset.

Subgroups of patients with young-onset type 2 diabetes in India reveal insulin deficiency as a major driver

Aim/hypothesisFive subgroups were described in European diabetes patients using a data driven machine learning approach on commonly measured variables. We aimed to test the applicability of this phenotyping in Indian individuals with young-onset type 2 diabetes.MethodsWe applied the European-derived centroids to Indian individuals with type 2 diabetes diagnosed before 45 years of age from the WellGen cohort (n = 1612). We also applied de novo k-means clustering to the WellGen cohort to validate the subgroups. We then compared clinical and metabolic-endocrine characteristics and the complication rates between the subgroups. We also compared characteristics of the WellGen subgroups with those of two young European cohorts, ANDIS (n = 962) and DIREVA (n = 420). Subgroups were also assessed in two other Indian cohorts, Ahmedabad (n = 187) and PHENOEINDY-2 (n = 205).ResultsBoth Indian and European young-onset type 2 diabetes patients were predominantly classified into severe insulin-deficient (SIDD) and mild obesity-related (MOD) subgroups, while the severe insulin-resistant (SIRD) and mild age-related (MARD) subgroups were rare. In WellGen, SIDD (53%) was more common than MOD (38%), contrary to findings in Europeans (Swedish 26% vs 68%, Finnish 24% vs 71%, respectively). A higher proportion of SIDD compared with MOD was also seen in Ahmedabad (57% vs 33%) and in PHENOEINDY-2 (67% vs 23%). Both in Indians and Europeans, the SIDD subgroup was characterised by insulin deficiency and hyperglycaemia, MOD by obesity, SIRD by severe insulin resistance and MARD by mild metabolic-endocrine disturbances. In WellGen, nephropathy and retinopathy were more prevalent in SIDD compared with MOD while the latter had higher prevalence of neuropathy.Conclusions /interpretationOur data identified insulin deficiency as the major driver of type 2 diabetes in young Indians, unlike in young European individuals in whom obesity and insulin resistance predominate. Our results provide useful clues to pathophysiological mechanisms and susceptibility to complications in type 2 diabetes in the young Indian population and suggest a need to review management strategies.Graphical abstract

Data Driven Calibration and Control of Compact Lightweight Series Elastic Actuators for Robotic Exoskeleton Gloves.

The working principle of a SEA is based on using an elastic material connected serially to the mechanical power source to simulate the dynamic behavior of a human muscle. Due to weight and size limitations of a wearable robotic exoskeleton, the hardware design of the SEA is limited. Compact and lightweight SEAs usually have noisy signal output, and can easily be deformed. This paper uses a compact lightweight SEA designed for exoskeleton gloves to demonstrate immeasurable strain and friction force which can cause an average of 34.31% and maximum of 44.7% difference in force measurement on such SEAs. This paper proposes two data driven machine learning methods to accurately calibrate and control SEAs. The multi-layer perception (MLP) method can reduce the average force measurement error to 10.18% and maximum error to 29.13%. The surface fitting method (SF) method can reduce the average force measurement error to 8.06% and maximum error to 35.72%. In control experiments, the weighted MLP method achieves an average of 0.21N force control difference, and the SF method achieves an average of 0.29N force control difference on the finger tips of the exoskeleton glove.

MACHINE LEARNING: AN OVERVIEW

Given the tremendous availability of data and computer power, there is a resurgence of interest in using data driven machine learning methods to solve issues where traditional engineering solutions are hampered by modeling or algorithmic flaws. The purpose of this article is to provide a comprehensive review of machine learning, including its history, types, applications, limitations and future prospects. In addition to this, the article also discusses the main point of difference between the field of artificial intelligence and machine learning.

Transition Modeling for Low Pressure Turbines Using Computational Fluid Dynamics Driven Machine Learning

Existing Reynolds Averaged Navier–Stokes-based transition models do not accurately predict separation induced transition for low pressure turbines. Therefore, in this paper, a novel framework based on computational fluids dynamics (CFD) driven machine learning coupled with multi-expression and multi-objective optimization is explored to develop models which can improve the transition prediction for the T106A low pressure turbine at an isentropic exit Reynolds number of Re2is=100,000. Model formulations are proposed for the transfer and laminar eddy viscosity terms of the laminar kinetic energy transition model using seven non-dimensional pi groups. The multi-objective optimization approach makes use of cost functions based on the suction-side wall-shear stress and the pressure coefficient. A family of solutions is thus developed, whose performance is assessed using Pareto analysis and in terms of physical characteristics of separated-flow transition. Two models are found which bring the wall-shear stress profile in the separated region at least two times closer to the reference high-fidelity data than the baseline transition model. As these models are able to accurately predict the flow coming off the blade trailing edge, they are also able to significantly enhance the wake-mixing prediction over the baseline model. This is the first known study which makes use of ‘CFD-driven’ machine learning to enhance the transition prediction for a non-canonical flow.

Application of data driven machine learning approach for modelling of non-linear filtration through granular porous media

Modeling the relationship between volumetric flux (m3/s) and gradient of hydraulic head (m) is extremely challenging in case of non-linear filtration through porous packing. Due to the uncertainties associated with the definition and quantification of characteristic length and velocity, experimental, theoretical and numerical modelling approaches are not widely applicable. The machine learning algorithms have proven to be extremely useful for predicting similar situations when the physical process is too complex to understand. The performance of Artificial Neural Network (ANN), Random Forest (RF), and Boosted Tree methods have been investigated in the study for predicting the gradient of hydraulic head (target variable) in case of non-linear filtration through porous packing. Velocity of flow, media size, porosity, kinetic viscosity, and shape factor obtained from a wide range of reported data in the literature was used as input features. All three models were observed to predict the output values with significant accuracies (R2>0.90) for wide range of data obtained from different sources. A RF method based sensitivity analysis was performed to study the relative importance of different hydrological parameters over the target variable. These parameters in terms of a decreasing order was ranked as velocity, diameter, viscosity, porosity, and shape factor. These type of models can aid the researchers, planners, and designers to predict the hydraulic gradient or volumetric flux in multiple scenarios associated with non-linear filtration through porous media such as oil and gas exploration wells, water filters, rockfill dams etc.

Which image-based phenotypes are most promising for using AI to understand cellular functions and why?

Which image-based phenotypes are most promising for using AI to understand cellular functions and why?

LSTM-based approach for predicting periodic motions of an impacting system via transient dynamics

Dynamically impacting systems are characterised with inherent instability and complex non-linear phenomena which makes it practically difficult to predict the steady state response of the system at transient periods. This study investigates the ability of a data driven machine learning method using Long Short-Term Memory networks to learn the complex nonlinearity associated with co-existing impact responses from limited transient data. A one-degree-of-freedom impact oscillator has been used to represent the bit–rock interaction for percussive drilling. Simulated data results show velocity measurements to contribute most to predicting steady state responses from transient dynamics with most of the network models reaching an accuracy of over 95%. Limitations to practically measurable variables in dynamic systems warranted the development of a feature based network model for impact motion classification. Experimental data from a two-degrees-of-freedom impacting system representing percussive bit penetration has been used to demonstrate the effectiveness of this method. The study thus provides a precise and less computational means of detecting and avoiding underperforming impact modes in percussive drilling.

Deep learning based automatic diagnosis of first-episode psychosis, bipolar disorder and healthy controls.

Neuroimaging data driven machine learning based predictive modeling and pattern recognition has been attracted strongly attention in biomedical sciences. Machine learning based diagnosis techniques are widely applied in diagnosis of neurological diseases. However, machine learning techniques are difficult to effectively extract deep information in neuroimaging data, resulting in low classification accuracy of mental illnesses. To address this problem, we propose a deep learning based automatic diagnosis first-episode psychosis (FEP), bipolar disorder (BD) and healthy controls (HC) method. Specifically, we design a convolutional neural network (CNN) framework to automatically diagnosis based on structural magnetic functional imaging (sMRI). Our dataset consists of 89 FEP patients, 40 BD patients and 83 HC. A three-way classifier (FEP vs. BD vs. HC) and three binary classifiers (FEP vs. BD, FEP vs. HC, BD vs. HC) are trained based on their gray matter volume images. Experiment results show that the performance of CNN-based method outperforms the classic classifiers both in two and three categories classification task. Our research reveals that abnormal gray matter volume is one of the main characteristics for discriminating FEP, BD and HC.

Application of data mining and machine learning in management accounting information system

ABSTRACT In the traditional management accounting information processing method, the method used to solve the problem is often fixed due to excessive assumptions. In order to improve its operating efficiency, combined with artificial intelligence information technology, this paper uses data mining algorithms to conduct data acquisition and rule exploration. Moreover, this paper uses statistics, machine learning and other techniques to analyze the correlation between attribute values and transform data into knowledge needed for decision-making. In addition, this paper combines machine learning algorithms to build an intelligent management accounting information system and realizes the close connection between corporate finance and business, which helps to form a closed-loop management between financial analysis, risk management, performance management, and decision-making. Finally, this paper designs experiments to verify the performance of the model. The research results show that the system constructed in this paper satisfies the intelligent demand of accounting information. REFERENCES [1] Jie Cai, Jiawei Luo, Shulin Wang, and Sheng Yang. Feature selection in machine learning: A new perspective. Neurocomputing, 300:70–79, 2018. [2] J. N. Goetz, A Brenning, H Petschko, and P. Leopold. Evaluating machine learning and statistical prediction techniques for landslide susceptibility modeling. Computers and Geosciences, 81:1–11, 2015. [3] Hamid Darabi, Bahram Choubin, Omid Rahmati, Ali Torabi Haghighi, Biswajeet Pradhan, and Bjorn Klove. Urban flood risk mapping using the GARP and QUEST models: A comparative study of machine learning techniques. Journal of Hydrology, 569:142–154, 2019. [4] Alvin Rajkomar, Jeffrey Dean, and Isaac Kohane. Machine learning in medicine. England Journal of Medicine, 380(14):1347–58, apr 2019. [5] Yang Xin, Lingshuang Kong, Zhi Liu, Yuling Chen, Yanmiao Li, Hongliang Zhu, Mingcheng Gao, Haixia Hou, and Chunhua Wang. Machine Learning and Deep Learning Methods for Cybersecurity. IEEE Access, 6:35365–35381, 2018. [6] Logan Ward, Ankit Agrawal, Alok Choudhary, and Christopher Wolverton. A general-purpose machinelearning framework for predicting properties of inorganic materials. npj Computational Materials, 2, 2016. [7] Puyu Feng, Bin Wang, De Li Liu, Cathy Waters, and Qiang Yu. Incorporating machine learning with biophysical model can improve the evaluation of climate extremes impacts on wheat yield in south-eastern Australia. Agricultural and Forest Meteorology, 275:100–113, 2019. [8] Konstantina Kourou, Themis P. Exarchos, Konstantinos P. Exarchos, Michalis V. Karamouzis, and DimitriosI. Fotiadis. Machine learning applications in cancer prognosis and prediction, 2015. [9] Saleema Amershi, Maya Cakmak, W. Bradley Knox, and Todd Kulesza. Power to the people: The role of humans in interactive machine learning. AI Magazine, 35(4):105–120, 2014. [10] V Rodriguez-Galiano, M. Sanchez-Castillo, M. Chica- Olmo, and M. Chica-Rivas. Machine learning predictive models for mineral prospectivity: An evaluation of neural networks, random forest, regression trees and support vector machines. Ore Geology Reviews, 71:804–818, 2015. [11] ConnorW Coley, Regina Barzilay, Tommi S Jaakkola, William H Green, and Klavs F Jensen. Prediction of Organic Reaction Outcomes Using Machine Learning. ACS Central Science, 3(5):434–443, may 2017. [12] Aritra Chowdhury, Elizabeth Kautz, Bulent Yener, and Daniel Lewis. Image driven machine learning methods for microstructure recognition. Computational MaterialsScience, 123:176–187, 2016.

Assessing temporal data partitioning scenarios for estimating reference evapotranspiration with machine learning techniques in arid regions

Recently, data driven machine learning techniques has been widely applied for modeling reference evapotranspiration (ETo) values under various climatic conditions taking into account the different number of sites and available data length. A major issue with applying those models is the proper selection of training/testing data sets. Although some spatial generalization approaches have been recommended for this purpose, there are no specified recommended local (temporal) data partitioning strategies for machine learning based ETo estimation. The present study evaluates different hold-out and k-fold validation temporal data partitioning strategies when using gene expression programming (GEP) technique to estimate daily ETo in arid regions. The k-fold validation strategies considered annual, monthly and growing season period patterns as test data sets. Although commonly used partitioning of the available patterns into training and testing sets gave accurate results, statistical analysis showed that the results obtained through k-fold validation assessment were more reliable. A two-block partitioning strategy with chronologic data selection for training and testing provided the most accurate results among the hold-out procedures (mean scatter index (SI) value of 0.162). Fixing the extreme ETo values as training data set in hold-out procedures provided the less accurate results with considerable over/underestimation of the ETo values (mean SI value was 0.506). Results on the basis of hold-out approaches can be biased or only partially valid depending on selection of the test data from the time series. K-fold validation yielded the lowest over/underestimations of ETo values. Further, considering monthly patterns as minimum affordable test size produced higher error magnitudes among k-fold validation strategies, while considering the complete patterns of one growing season provided more accurate results among k-fold validation strategies.

Supervised Learning in All FeFET-Based Spiking Neural Network: Opportunities and Challenges.

The two possible pathways toward artificial intelligence (AI)—(i) neuroscience-oriented neuromorphic computing [like spiking neural network (SNN)] and (ii) computer science driven machine learning (like deep learning) differ widely in their fundamental formalism and coding schemes (Pei et al., 2019). Deviating from traditional deep learning approach of relying on neuronal models with static nonlinearities, SNNs attempt to capture brain-like features like computation using spikes. This holds the promise of improving the energy efficiency of the computing platforms. In order to achieve a much higher areal and energy efficiency compared to today’s hardware implementation of SNN, we need to go beyond the traditional route of relying on CMOS-based digital or mixed-signal neuronal circuits and segregation of computation and memory under the von Neumann architecture. Recently, ferroelectric field-effect transistors (FeFETs) are being explored as a promising alternative for building neuromorphic hardware by utilizing their non-volatile nature and rich polarization switching dynamics. In this work, we propose an all FeFET-based SNN hardware that allows low-power spike-based information processing and co-localized memory and computing (a.k.a. in-memory computing). We experimentally demonstrate the essential neuronal and synaptic dynamics in a 28 nm high-K metal gate FeFET technology. Furthermore, drawing inspiration from the traditional machine learning approach of optimizing a cost function to adjust the synaptic weights, we implement a surrogate gradient (SG) learning algorithm on our SNN platform that allows us to perform supervised learning on MNIST dataset. As such, we provide a pathway toward building energy-efficient neuromorphic hardware that can support traditional machine learning algorithms. Finally, we undertake synergistic device-algorithm co-design by accounting for the impacts of device-level variation (stochasticity) and limited bit precision of on-chip synaptic weights (available analog states) on the classification accuracy.

Hybrid wireless aided volunteer computing paradigm

Big-data acquisition and processing is important in developing value driven machine learning applications. This is challenging in compute resource-constrained scenarios. Compute resource-constrained scenarios arise due to low capacity of installed cloud infrastructure and low availability of high speed internet links. These factors limit the ability to process crowd-sourced data to develop machine learning applications. The volunteer computing paradigm is found to be suitable for addressing these challenges. Volunteer computing paradigm makes use of computing nodes provided by users distributed over a geographical area. It leverages on the availability of volunteers with low cost computing entities. This paper proposes the fractionated computing system (FCS) to address the challenges described above. FCS incorporates intelligent compute node selection and uses high performance end-user computing nodes (laptops) to process the crowd-sourced data. The performance of FCS is investigated against the existing method of using cloud servers. Results show that FCS reduces acquisition costs and power consumption by 35% and up to 56.5% on average, respectively. The watt per bit expended in processing crowd-sourced data is also enhanced by up to 98% on average. In addition, the use of FCS enhances memory resources accessible for data processing. Simulations show that increasing memory in modular computing entities by up to 58.7% enhances memory available across network of modular volunteer computing nodes by 0.5 EB. The use of end-user nodes with modular communication subsystems instead of end-user computing nodes with non-modular communication sub-systems enhances channel capacity by 37.5% on average.

Estimation of high energy steam piping degradation using hybrid recurrent neural networks

The degradation of high energy piping systems is very complex to simulate due to the many variables that influence the useful lives of such systems. Estimation of the extent of degradation is however important in the maintenance planning process. In this research the use of data driven machine learning techniques to deal with this complex problem is investigated. A hybrid recurrent neural network is created that consists of a combined recurrent neural network and a feed forward neural network. The hybrid model is trained on historical data that has been captured over a six-year time period. The following variables related to piping system components are used as inputs to the machine learning model: operating temperature and pressure time history, distance to the closest anchor point, distances to neighbouring supports, elevation survey readings, as well as the last known creep damage measurements on the component. The model is created in Python using the Tensorflow library. A recurrent neural networks (RNN) is employed, namely the gated recurrent unit (GRU). The adaptive movement estimation optimization algorithm, called Adam, is used to optimize the machine learning model. The trained model is able to predict the degradation classification of a component with an accuracy of up to 92% on the training dataset and up to 55% on the validation data set. When using this model to predict components with high creep damage, more than 400 voids per mm2 a hit rate of 25% is achieved. The current system employed at operating power stations shows a historic hit rate of 14%. This is a significant increase in performance and could be used to compile more efficient inspection plans. The model is successful in recognising patterns within the data and offers an automated way to parse large data sets that consist of a temporal and static data mixture simultaneously. Conventional data driven models are only able to look at either temporal data or static data. This suggests a generic approach to make objective decisions on similar complex data driven problems and its application is not limited to this particular problem. The methods applied in this research are expected to perform even better on problems where the frequency of data collection is higher than what is used in this research.

Integrated data and knowledge driven methodology for human activity recognition

Human activity recognition has been a popular research area concerned with identifying the specific movement or action of a person based on variety of sensor data. Conventional human activity recognition approaches are mainly data driven, which are not working well for composite activity recognition due to the complexity and uncertainty of real scenarios. We propose in this paper a hierarchical structure-based framework and methodology for human activity recognition by an integration of data-driven approach and knowledge-based approach, which provides an interesting framework capable of bridging lower-level pattern recognition and higher-level knowledge for reasoning and explanation. More specifically, this approach constructs a hierarchical structure for representing the composite activity by a composition of lower-level actions and gestures according to its semantic meaning. This hierarchical structure is then transformed into formal syntactical logical formulas and rules, based on which the resolution based automated reasoning is applied to recognize the composite activity given the recognized lower-level actions by using data driven machine learning methods. The work is the validated using some open-source data about video based human activity recognition. The present work provides a promising framework and application illustration of integration of machine learning and symbolic reasoning.