scholarly journals A Responsible Machine Learning Workflow with Focus on Interpretable Models, Post-hoc Explanation, and Discrimination Testing

Information ◽  
2020 ◽  
Vol 11 (3) ◽  
pp. 137
Author(s):  
Navdeep Gill ◽  
Patrick Hall ◽  
Kim Montgomery ◽  
Nicholas Schmidt
Keyword(s):  
Machine Learning ◽  
Simulated Data ◽  
Gradient Boosting ◽  
Marginal Effect ◽  
Adverse Action ◽  
Discrimination Testing ◽  
Interpretable Models ◽  
Viable Approach ◽  
Constrained Models ◽  
Post Hoc

This manuscript outlines a viable approach for training and evaluating machine learning systems for high-stakes, human-centered, or regulated applications using common Python programming tools. The accuracy and intrinsic interpretability of two types of constrained models, monotonic gradient boosting machines and explainable neural networks, a deep learning architecture well-suited for structured data, are assessed on simulated data and publicly available mortgage data. For maximum transparency and the potential generation of personalized adverse action notices, the constrained models are analyzed using post-hoc explanation techniques including plots of partial dependence and individual conditional expectation and with global and local Shapley feature importance. The constrained model predictions are also tested for disparate impact and other types of discrimination using measures with long-standing legal precedents, adverse impact ratio, marginal effect, and standardized mean difference, along with straightforward group fairness measures. By combining interpretable models, post-hoc explanations, and discrimination testing with accessible software tools, this text aims to provide a template workflow for machine learning applications that require high accuracy and interpretability and that mitigate risks of discrimination.

scholarly journals EXPLAIN IT TO ME – FACING REMOTE SENSING CHALLENGES IN THE BIO- AND GEOSCIENCES WITH EXPLAINABLE MACHINE LEARNING

2020 ◽  
Vol V-3-2020 ◽  
pp. 817-824
Author(s):  
R. Roscher ◽  
B. Bohn ◽  
M. F. Duarte ◽  
J. Garcke
Keyword(s):  
Machine Learning ◽  
Remote Sensing ◽  
Neural Networks ◽  
Simulated Data ◽  
High Accuracy ◽  
Learning Approaches ◽  
Learning Methods ◽  
Machine Learning Methods ◽  
Recent Advances ◽  
Post Hoc

Abstract. For some time now, machine learning methods have been indispensable in many application areas. Especially with the recent development of efficient neural networks, these methods are increasingly used in the sciences to obtain scientific outcomes from observational or simulated data. Besides a high accuracy, a desired goal is to learn explainable models. In order to reach this goal and obtain explanation, knowledge from the respective domain is necessary, which can be integrated into the model or applied post-hoc. We discuss explainable machine learning approaches which are used to tackle common challenges in the bio- and geosciences, such as limited amount of labeled data or the provision of reliable and scientific consistent results. We show that recent advances in machine learning to enhance transparency, interpretability, and explainability are helpful in overcoming these challenges.

scholarly journals Controlling for effects of confounding variables on machine learning predictions

2020 ◽  
Author(s):  
Richard Dinga ◽  
Lianne Schmaal ◽  
Brenda W.J.H. Penninx ◽  
Dick J. Veltman ◽  
Andre F. Marquand
Keyword(s):  
Machine Learning ◽  
Predictive Models ◽  
Simulated Data ◽  
Predictive Performance ◽  
Head Motion ◽  
Learning Models ◽  
Confounding Variables ◽  
And Performance ◽  
Post Hoc ◽  
Machine Learning Models

ABSTRACTMachine learning predictive models are being used in neuroimaging to predict information about the task or stimuli or to identify potentially clinically useful biomarkers. However, the predictions can be driven by confounding variables unrelated to the signal of interest, such as scanner effect or head motion, limiting the clinical usefulness and interpretation of machine learning models. The most common method to control for confounding effects is regressing out the confounding variables separately from each input variable before machine learning modeling. However, we show that this method is insufficient because machine learning models can learn information from the data that cannot be regressed out. Instead of regressing out confounding effects from each input variable, we propose controlling for confounds post-hoc on the level of machine learning predictions. This allows partitioning of the predictive performance into the performance that can be explained by confounds and performance independent of confounds. This approach is flexible and allows for parametric and non-parametric confound adjustment. We show in real and simulated data that this method correctly controls for confounding effects even when traditional input variable adjustment produces false-positive findings.

scholarly journals The accuracy versus interpretability trade-off in fraud detection model

Data & Policy ◽  
2021 ◽  
Vol 3 ◽  
Author(s):  
Anna Nesvijevskaia ◽  
Sophie Ouillade ◽  
Pauline Guilmin ◽  
Jean-Daniel Zucker
Keyword(s):  
Machine Learning ◽  
Concept Drift ◽  
Fraud Detection ◽  
Learning Approaches ◽  
Policy Makers ◽  
Trade Off ◽  
Detection Model ◽  
Hybrid Approaches ◽  
Interpretable Models ◽  
Post Hoc

Abstract Like a hydra, fraudsters adapt and circumvent increasingly sophisticated barriers erected by public or private institutions. Among these institutions, banks must quickly take measures to avoid losses while guaranteeing the satisfaction of law-abiding customers. Facing an expanding flow of operations, effective banking relies on data analytics to support established risk control processes, but also on a better understanding of the underlying fraud mechanism. In addition, fraud being a criminal offence, the evidential aspect of the process must also be considered. These legal, operational, and strategic constraints lead to compromises on the means to be implemented for fraud management. This paper first focuses on the translation of practical questions raised in the banking industry at each step of the fraud management process into performance evaluation required to design a fraud detection model. Secondly, it considers a range of machine learning approaches that address these specificities: the imbalance between fraudulent and nonfraudulent operations, the lack of fully trusted labels, the concept-drift phenomenon, and the unavoidable trade-off between accuracy and interpretability of detection. This state-of-the-art review sheds some light on a technology race between black box machine learning models improved by post-hoc interpretation and intrinsic interpretable models boosted to gain accuracy. Finally, it discusses how concrete and promising hybrid approaches can provide pragmatic, short-term answers to banks and policy makers without swallowing up stakeholders with economical and ethical stakes in this technological race.

scholarly journals Comparison of Random Forest and Gradient Boosting Fingerprints to Enhance an Outdoor Radio-frequency Localization System

2020 ◽  
Author(s):  
Marcelo Nogueira de Sousa ◽  
Ricardo Sant'Ana ◽  
Riegel P. Fernandes ◽  
Julio Cesár Duarte ◽  
José A. Aploinário ◽  
...  
Keyword(s):  
Machine Learning ◽  
Random Forest ◽  
Radio Frequency ◽  
Simulated Data ◽  
Gradient Boosting ◽  
Learning Tools ◽  
Learning Framework ◽  
Error Mitigation ◽  
High Bias ◽  
Outdoor Localization

Abstract Machine Learning framework adds a new dimension to the localization estimation problem; it tries to find the most likely position using processed features in a radio map. This paper compares the performance of two machine learning tools, Random Forest (RF) and XGBoost, in exploiting the multipath information for outdoor localization problem. The investigation was carried out in a noisy outdoor scenario, where non-line-of-sight between target and sensors may affect the location of a radio-frequency emitter strongly. It is possible to improve the position system performance by using fingerprints techniques that employ multipath information in a Machine Learning framework, which operate a dataset generated by ray-tracing simulation. Usually, real measurements produce the fingerprints localization features, and there is mismatching with the simulated data. Another drawback of NLOS features extraction is the noise level that occurs in position processing. Random Forest algorithm uses fully grown decision trees to classify possible emitter position, trying to achieve error mitigation by reducing variance. On the other hand, XGBoost approach uses weak learners, defined by high bias and low variance. The results of the simulation performed aims to be used as a design parameter to perform hyperparameter refinements in similar multipath localization problems.

Forecasting US movies box office performances in Turkey using machine learning algorithms

2020 ◽  
Vol 39 (5) ◽  
pp. 6579-6590
Author(s):  
Sandy Çağlıyor ◽  
Başar Öztayşi ◽  
Selime Sezgin
Keyword(s):  
Machine Learning ◽  
Global Economy ◽  
Learning Algorithms ◽  
Forecast Model ◽  
Machine Learning Algorithms ◽  
Gradient Boosting ◽  
High Stakes ◽  
Box Office ◽  
Industry Forecast ◽  
The Impact

The motion picture industry is one of the largest industries worldwide and has significant importance in the global economy. Considering the high stakes and high risks in the industry, forecast models and decision support systems are gaining importance. Several attempts have been made to estimate the theatrical performance of a movie before or at the early stages of its release. Nevertheless, these models are mostly used for predicting domestic performances and the industry still struggles to predict box office performances in overseas markets. In this study, the aim is to design a forecast model using different machine learning algorithms to estimate the theatrical success of US movies in Turkey. From various sources, a dataset of 1559 movies is constructed. Firstly, independent variables are grouped as pre-release, distributor type, and international distribution based on their characteristic. The number of attendances is discretized into three classes. Four popular machine learning algorithms, artificial neural networks, decision tree regression and gradient boosting tree and random forest are employed, and the impact of each group is observed by compared by the performance models. Then the number of target classes is increased into five and eight and results are compared with the previously developed models in the literature.

In silico Prediction of Inhibitory Constant of Thrombin Inhibitors Using Machine Learning

2019 ◽  
Vol 21 (9) ◽  
pp. 662-669 ◽  
Author(s):  
Junnan Zhao ◽  
Lu Zhu ◽  
Weineng Zhou ◽  
Lingfeng Yin ◽  
Yuchen Wang ◽  
...  
Keyword(s):  
Machine Learning ◽  
Prediction Models ◽  
Regression Tree ◽  
Large Data ◽  
Thrombin Inhibitors ◽  
Coagulation Cascade ◽  
Gradient Boosting ◽  
Support Vector ◽  
Data Set ◽  
Descriptor Selection

Background: Thrombin is the central protease of the vertebrate blood coagulation cascade, which is closely related to cardiovascular diseases. The inhibitory constant Ki is the most significant property of thrombin inhibitors. Method: This study was carried out to predict Ki values of thrombin inhibitors based on a large data set by using machine learning methods. Taking advantage of finding non-intuitive regularities on high-dimensional datasets, machine learning can be used to build effective predictive models. A total of 6554 descriptors for each compound were collected and an efficient descriptor selection method was chosen to find the appropriate descriptors. Four different methods including multiple linear regression (MLR), K Nearest Neighbors (KNN), Gradient Boosting Regression Tree (GBRT) and Support Vector Machine (SVM) were implemented to build prediction models with these selected descriptors. Results: The SVM model was the best one among these methods with R2=0.84, MSE=0.55 for the training set and R2=0.83, MSE=0.56 for the test set. Several validation methods such as yrandomization test and applicability domain evaluation, were adopted to assess the robustness and generalization ability of the model. The final model shows excellent stability and predictive ability and can be employed for rapid estimation of the inhibitory constant, which is full of help for designing novel thrombin inhibitors.

A Study on Host Tropism Determinants of Influenza Virus Using Machine Learning

2020 ◽  
Vol 15 (2) ◽  
pp. 121-134 ◽  
Author(s):  
Eunmi Kwon ◽  
Myeongji Cho ◽  
Hayeon Kim ◽  
Hyeon S. Son
Keyword(s):  
Machine Learning ◽  
Amino Acids ◽  
Influenza Virus ◽  
Random Forest ◽  
Physicochemical Properties ◽  
Protein Sequences ◽  
Influenza Viruses ◽  
Host Tropism ◽  
Post Hoc ◽  
Ha Protein

Background: The host tropism determinants of influenza virus, which cause changes in the host range and increase the likelihood of interaction with specific hosts, are critical for understanding the infection and propagation of the virus in diverse host species. Methods: Six types of protein sequences of influenza viral strains isolated from three classes of hosts (avian, human, and swine) were obtained. Random forest, naïve Bayes classification, and knearest neighbor algorithms were used for host classification. The Java language was used for sequence analysis programming and identifying host-specific position markers. Results: A machine learning technique was explored to derive the physicochemical properties of amino acids used in host classification and prediction. HA protein was found to play the most important role in determining host tropism of the influenza virus, and the random forest method yielded the highest accuracy in host prediction. Conserved amino acids that exhibited host-specific differences were also selected and verified, and they were found to be useful position markers for host classification. Finally, ANOVA analysis and post-hoc testing revealed that the physicochemical properties of amino acids, comprising protein sequences combined with position markers, differed significantly among hosts. Conclusion: The host tropism determinants and position markers described in this study can be used in related research to classify, identify, and predict the hosts of influenza viruses that are currently susceptible or likely to be infected in the future.

Predicting Undesired Treatment Outcome in Mental Healthcare: Machine Learning Study (Preprint)

2019 ◽  
Author(s):  
Kasper Van Mens ◽  
Joran Lokkerbol ◽  
Richard Janssen ◽  
Robert de Lange ◽  
Bea Tiemens
Keyword(s):  
Machine Learning ◽  
Treatment Outcome ◽  
Mental Health Treatment ◽  
Mental Healthcare ◽  
Machine Learning Algorithms ◽  
Gradient Boosting ◽  
Trade Off ◽  
Trade Offs ◽  
Outcome Monitoring ◽  
Extreme Gradient Boosting

BACKGROUND It remains a challenge to predict which treatment will work for which patient in mental healthcare. OBJECTIVE In this study we compare machine algorithms to predict during treatment which patients will not benefit from brief mental health treatment and present trade-offs that must be considered before an algorithm can be used in clinical practice. METHODS Using an anonymized dataset containing routine outcome monitoring data from a mental healthcare organization in the Netherlands (n = 2,655), we applied three machine learning algorithms to predict treatment outcome. The algorithms were internally validated with cross-validation on a training sample (n = 1,860) and externally validated on an unseen test sample (n = 795). RESULTS The performance of the three algorithms did not significantly differ on the test set. With a default classification cut-off at 0.5 predicted probability, the extreme gradient boosting algorithm showed the highest positive predictive value (ppv) of 0.71(0.61 – 0.77) with a sensitivity of 0.35 (0.29 – 0.41) and area under the curve of 0.78. A trade-off can be made between ppv and sensitivity by choosing different cut-off probabilities. With a cut-off at 0.63, the ppv increased to 0.87 and the sensitivity dropped to 0.17. With a cut-off of at 0.38, the ppv decreased to 0.61 and the sensitivity increased to 0.57. CONCLUSIONS Machine learning can be used to predict treatment outcomes based on routine monitoring data.This allows practitioners to choose their own trade-off between being selective and more certain versus inclusive and less certain.

scholarly journals Evaluation of Three Different Machine Learning Methods for Object-Based Artificial Terrace Mapping—A Case Study of the Loess Plateau, China

2021 ◽  
Vol 13 (5) ◽  
pp. 1021
Author(s):  
Hu Ding ◽  
Jiaming Na ◽  
Shangjing Jiang ◽  
Jie Zhu ◽  
Kai Liu ◽  
...  
Keyword(s):  
Machine Learning ◽  
Random Forest ◽  
Loess Plateau ◽  
Water Conservation ◽  
Nearest Neighbor ◽  
Gradient Boosting ◽  
K Nearest Neighbor ◽  
The Loess Plateau ◽  
Object Based ◽  
Extreme Gradient Boosting

Artificial terraces are of great importance for agricultural production and soil and water conservation. Automatic high-accuracy mapping of artificial terraces is the basis of monitoring and related studies. Previous research achieved artificial terrace mapping based on high-resolution digital elevation models (DEMs) or imagery. As a result of the importance of the contextual information for terrace mapping, object-based image analysis (OBIA) combined with machine learning (ML) technologies are widely used. However, the selection of an appropriate classifier is of great importance for the terrace mapping task. In this study, the performance of an integrated framework using OBIA and ML for terrace mapping was tested. A catchment, Zhifanggou, in the Loess Plateau, China, was used as the study area. First, optimized image segmentation was conducted. Then, features from the DEMs and imagery were extracted, and the correlations between the features were analyzed and ranked for classification. Finally, three different commonly-used ML classifiers, namely, extreme gradient boosting (XGBoost), random forest (RF), and k-nearest neighbor (KNN), were used for terrace mapping. The comparison with the ground truth, as delineated by field survey, indicated that random forest performed best, with a 95.60% overall accuracy (followed by 94.16% and 92.33% for XGBoost and KNN, respectively). The influence of class imbalance and feature selection is discussed. This work provides a credible framework for mapping artificial terraces.

Sign in / Sign up

Export Citation Format

Share Document