Complex Machine Learning Research Articles

Enhancing trust and interpretability of complex machine learning models using local interpretable model agnostic shap explanations

Enhancing trust and interpretability of complex machine learning models using local interpretable model agnostic shap explanations

A new model for counterfactual analysis for functional data

AbstractCounterfactual explanations have become a very popular interpretability tool to understand and explain how complex machine learning models make decisions for individual instances. Most of the research on counterfactual explainability focuses on tabular and image data and much less on models dealing with functional data. In this paper, a counterfactual analysis for functional data is addressed, in which the goal is to identify the samples of the dataset from which the counterfactual explanation is made of, as well as how they are combined so that the individual instance and its counterfactual are as close as possible. Our methodology can be used with different distance measures for multivariate functional data and is applicable to any score-based classifier. We illustrate our methodology using two different real-world datasets, one univariate and another multivariate.

Rationalizing general limitations in assessing and comparing methods for compound potency prediction

Compound potency predictions play a major role in computational drug discovery. Predictive methods are typically evaluated and compared in benchmark calculations that are widely applied. Previous studies have revealed intrinsic limitations of potency prediction benchmarks including very similar performance of increasingly complex machine learning methods and simple controls and narrow error margins separating machine learning from randomized predictions. However, origins of these limitations are currently unknown. We have carried out an in-depth analysis of potential reasons leading to artificial outcomes of potency predictions using different methods. Potency predictions on activity classes typically used in benchmark settings were found to be determined by compounds with intermediate potency close to median values of the compound data sets. The potency of these compounds was consistently predicted with high accuracy, without the need for learning, which dominated the results of benchmark calculations, regardless of the activity classes used. Taken together, our findings provide a clear rationale for general limitations of compound potency benchmark predictions and a basis for the design of alternative test systems for methodological comparisons.

Unifying complexity science and machine learning

Complexity science and machine learning are two complementary approaches to discovering and encoding regularities in irreducibly high dimensional phenomena. Whereas complexity science represents a coarse-grained paradigm of understanding, machine learning is a fine-grained paradigm of prediction. Both approaches seek to solve the “Wigner-Reversal” or the unreasonable ineffectiveness of mathematics in the adaptive domain where broken symmetries and broken ergodicity dominate. In order to integrate these paradigms I introduce the idea of “Meta-Ockham” which 1) moves minimality from the description of a model for a phenomenon to a description of a process for generating a model and 2) describes low dimensional features–schema–in these models. Reinforcement learning and natural selection are both parsimonious in this revised sense of minimal processes that parameterize arbitrarily high-dimensional inductive models containing latent, low-dimensional, regularities. I describe these models as “super-Humean” and discuss the scientic value of analyzing their latent dimensions as encoding functional schema.

Assessing growth potential of careers with occupational mobility network and ensemble framework

The growth potential of a career reflects its future prospects and is an important consideration for individuals and organizations when career planning. There is still a lack of quantitative assessment tools for growth potential of careers. In this study, considering the key role of human capital in human resource management, as well as the excellent performance of complex network and machine learning in big data analysis and prediction, a career growth potential assessment model with human capital ensemble is proposed through human capital-based occupational mobility network and ensemble learning. First, an occupational mobility network is constructed based on online professional dataset to associate occupations with each other. Then, five dimensions of human capital measurements are designed to quantify human capital in terms of education, experience, social capital, occupational size, and concentration. These are then combined with the occupational mobility network to create a new network that depicts human capital flows among occupations. Finally, an ensemble framework for assessing career growth potential is constructed to integrate multidimensional human capital information in the network and obtain quantitative scores of growth potential. This study is the original attempt to adopt a data-driven idea and an intelligent approach to understand career growth potential. The experimental results show that it also makes a useful exploration for modeling human capital flows and intelligent assessment of career prospects.

Scientific Machine Learning of 2D Perovskite Nanosheet Formation.

We apply a scientific machine learning (ML) framework to aid the prediction and understanding of nanomaterial formation processes via a joint spectral-kinetic model. We apply this framework to study the nucleation and growth of two-dimensional (2D) perovskite nanosheets. Colloidal nanomaterials have size-dependent optical properties and can be observed in situ, all of which make them a good model for understanding the complex processes of nucleation, growth, and phase transformation of 2D perovskites. Our results demonstrate that this model nanomaterial can form through two processes at the nanoscale: either via a layer-by-layer chemical exfoliation process from lead bromide nanocrystals or via direct nucleation from precursors. We utilize a phenomenological kinetic analysis to study the exfoliation process and scientific machine learning to study the direct nucleation and growth and discuss the circumstances under which it is more appropriate to use phenomenological or more complex machine learning models. Data for both analysis techniques are collected through in situ spectroscopy in a stopped flow chamber, incorporating over 500,000 spectra taken under more than 100 different conditions. More broadly, our research shows that the ability to utilize and integrate traditional kinetics and machine learning methods will greatly assist in the understanding of complex chemical systems.

Explainable Machine Learning for Credit Risk Management When Features are Dependent

ABSTRACT Complex Machine Learning (ML) models used to support decision-making in peer-to-peer (P2P) lending often lack clear, accurate, and interpretable explanations. While the game-theoretic concept of Shapley values and its computationally efficient variant Kernel SHAP may be employed for this aim, similarly to other existing methods, the latter makes the assumption that the features are independent. The assumption of uncorrelated features in credit risk management is fairly restrictive and, thus, prediction explanations coming from correlated features might result in highly misleading Shapley values, even when considering simple models. We therefore propose an evaluation of different dependent-feature estimation methods of Kernel SHAP for classification purposes in credit risk management. We show that dependent-feature estimation of Shapley values can improve the understanding of true prediction explanations, their robustness and is essential for better identifying the most relevant variables to default predictions coming from black-box ML models.

Machine Learning Method for Changepoint Detection in Short Time Series Data

Analysis of data is crucial in waste management to improve effective planning from both short- and long-term perspectives. Real-world data often presents anomalies, but in the waste management sector, anomaly detection is seldom performed. The main goal and contribution of this paper is a proposal of a complex machine learning framework for changepoint detection in a large number of short time series from waste management. In such a case, it is not possible to use only an expert-based approach due to the time-consuming nature of this process and subjectivity. The proposed framework consists of two steps: (1) outlier detection via outlier test for trend-adjusted data, and (2) changepoints are identified via comparison of linear model parameters. In order to use the proposed method, it is necessary to have a sufficient number of experts’ assessments of the presence of anomalies in time series. The proposed framework is demonstrated on waste management data from the Czech Republic. It is observed that certain waste categories in specific regions frequently exhibit changepoints. On the micro-regional level, approximately 31.1% of time series contain at least one outlier and 16.4% exhibit changepoints. Certain groups of waste are more prone to the occurrence of anomalies. The results indicate that even in the case of aggregated data, anomalies are not rare, and their presence should always be checked.

Biomarkers in the Pathogenesis, Diagnosis, and Treatment of Systemic Sclerosis.

Systemic sclerosis (SSc) is a complex autoimmune disease characterized by vascular damage, vasoinstability, and decreased perfusion with ischemia, inflammation, and exuberant fibrosis of the skin and internal organs. Biomarkers are analytic indicators of the biological and disease processes within an individual that can be accurately and reproducibly measured. The field of biomarkers in SSc is complex as recent studies have implicated at least 240 pathways and dysregulated proteins in SSc pathogenesis. Anti-nuclear antibodies (ANA) are classical biomarkers with well-described clinical classifications and are present in more than 90% of SSc patients and include anti-centromere, anti-Th/To, anti-RNA polymerase III, and anti-topoisomerase I antibodies. Transforming growth factor-β (TGF-β) is central to the fibrotic process of SSc and is intimately intertwined with other biomarkers. Tyrosine kinases, interferon-1 signaling, IL-6 signaling, endogenous thrombin, peroxisome proliferator-activated receptors (PPARs), lysophosphatidic acid receptors, and amino acid metabolites are new biomarkers with the potential for developing new therapeutic agents. Other biomarkers implicated in SSc-ILD include signal transducer and activator of transcription 4 (STAT4), CD226 (DNAX accessory molecule 1), interferon regulatory factor 5 (IRF5), interleukin-1 receptor-associated kinase-1 (IRAK1), connective tissue growth factor (CTGF), pyrin domain containing 1 (NLRP1), T-cell surface glycoprotein zeta chain (CD3ζ) or CD247, the NLR family, SP-D (surfactant protein), KL-6, leucine-rich α2-glycoprotein-1 (LRG1), CCL19, genetic factors including DRB1 alleles, the interleukins (IL-1, IL-4, IL-6, IL-8, IL-10 IL-13, IL-16, IL-17, IL-18, IL-22, IL-32, and IL-35), the chemokines CCL (2,3,5,13,20,21,23), CXC (8,9,10,11,16), CX3CL1 (fractalkine), and GDF15. Adiponectin (an indicator of PPAR activation) and maresin 1 are reduced in SSc patients. A new trend has been the use of biomarker panels with combined complex multifactor analysis, machine learning, and artificial intelligence to determine disease activity and response to therapy. The present review is an update of the various biomarker molecules, pathways, and receptors involved in the pathology of SSc.

Mathematical Modeling and Analysis of Credit Scoring Using the LIME Explainer: A Comprehensive Approach

Credit scoring models serve as pivotal instruments for lenders and financial institutions, facilitating the assessment of creditworthiness. Traditional models, while instrumental, grapple with challenges related to efficiency and subjectivity. The advent of machine learning heralds a transformative era, offering data-driven solutions that transcend these limitations. This research delves into a comprehensive analysis of various machine learning algorithms, emphasizing their mathematical underpinnings and their applicability in credit score classification. A comprehensive evaluation is conducted on a range of algorithms, including logistic regression, decision trees, support vector machines, and neural networks, using publicly available credit datasets. Within the research, a unified mathematical framework is introduced, which encompasses preprocessing techniques and critical algorithms such as Particle Swarm Optimization (PSO), the Light Gradient Boosting Model, and Extreme Gradient Boosting (XGB), among others. The focal point of the investigation is the LIME (Local Interpretable Model-agnostic Explanations) explainer. This study offers a comprehensive mathematical model using the LIME explainer, shedding light on its pivotal role in elucidating the intricacies of complex machine learning models. This study’s empirical findings offer compelling evidence of the efficacy of these methodologies in credit scoring, with notable accuracies of 88.84%, 78.30%, and 77.80% for the Australian, German, and South German datasets, respectively. In summation, this research not only amplifies the significance of machine learning in credit scoring but also accentuates the importance of mathematical modeling and the LIME explainer, providing a roadmap for practitioners to navigate the evolving landscape of credit assessment.

A k-additive Choquet integral-based approach to approximate the SHAP values for local interpretability in machine learning

Besides accuracy, recent studies on machine learning models have been addressing the question on how the obtained results can be interpreted. Indeed, while complex machine learning models are able to provide very good results in terms of accuracy even in challenging applications, it is difficult to interpret them. Aiming at providing some interpretability for such models, one of the most famous methods, called SHAP, borrows the Shapley value concept from game theory in order to locally explain the predicted outcome of an instance of interest. As the SHAP values calculation needs previous computations on all possible coalitions of attributes, its computational cost can be very high. Therefore, a SHAP-based method called Kernel SHAP adopts a strategy that approximates such values with less computational effort. However, we see two weaknesses in Kernel SHAP: its formulation is difficult to understand and it does not consider further game theory assumptions that could reduce the computational cost. Therefore, in this paper, we propose a novel approach that addresses such weaknesses. Firstly, we provide a straightforward formulation of a SHAP-based method for local interpretability by using the Choquet integral, which leads to both Shapley values and Shapley interaction indices. Thereafter, we propose to adopt the concept of k-additive games from game theory, which contributes to reduce the computational effort when estimating the SHAP values. The obtained results attest that our proposal needs less computations on coalitions of attributes to approximate the SHAP values.

Physics-informed Gaussian process for tool wear prediction

The tool wear monitoring (TWM) system plays an increasingly important role to ensure high quality finishing and system safety in advanced CNC machining process. The pure data-based TWM approaches generally needs to develop complex machine learning models and require massive sensory data to learn the models to reach high monitoring accuracy, while the physics-based tool wear models are simple but hard to adapt to varied working conditions. In order to incorporate the benefits of both methods, a novel physics-informed Gaussian process model is developed to predict the tool wear. Different from the traditional approaches, three tool wear physical models are introduced to develop the physics-informed Gaussian process regression (PB-GPR) model. The wear model is applied to constrain the mean function of the Gaussian process, so that the PB-GPR is more in line with the actual tool wear. At the same time, the model can initiate small data training to meet limited tool wear labels in practice, and then update the model with new measurements. Multi-sensor signals are collected and multi-domain features are extracted for the model learning. The proposed approach is validated from high speed milling experiments. The results show a significant performance improvement including tool wear prediction accuracy and robustness in extrapolation compared to the conventional machine learning methods.

Stock picking with machine learning

AbstractWe analyze machine learning algorithms for stock selection. Our study builds on weekly data for the historical constituents of the S&P500 over the period from January 1999 to March 2021 and builds on typical equity factors, additional firm fundamentals, and technical indicators. A variety of machine learning models are trained on the binary classification task to predict whether a specific stock outperforms or underperforms the cross‐sectional median return over the subsequent week. We analyze weekly trading strategies that invest in stocks with the highest predicted outperformance probability. Our empirical results show substantial and significant outperformance of machine learning‐based stock selection models compared to an equally weighted benchmark. Interestingly, we find more simplistic regularized logistic regression models to perform similarly well compared to more complex machine learning models. The results are robust when applied to the STOXX Europe 600 as alternative asset universe.

Machine Learning to Predict Apical Lesions: A Cross-Sectional and Model Development Study.

(1) Background: We aimed to identify factors associated with the presence of apical lesions (AL) in panoramic radiographs and to evaluate the predictive value of the identified factors. (2) Methodology: Panoramic radiographs from 1071 patients (age: 11-93 a, mean: 50.6 a ± 19.7 a) with 27,532 teeth were included. Each radiograph was independently assessed by five experienced dentists for AL. A range of shallow machine learning algorithms (logistic regression, k-nearest neighbor, decision tree, random forest, support vector machine, adaptive and gradient boosting) were employed to identify factors at both the patient and tooth level associated with AL and to predict AL. (3) Results: AL were detected in 522 patients (48.7%) and 1133 teeth (4.1%), whereas males showed a significantly higher prevalence than females (52.5%/44.8%; p < 0.05). Logistic regression found that an existing root canal treatment was the most important risk factor (adjusted Odds Ratio 16.89; 95% CI: 13.98-20.41), followed by the tooth type 'molar' (2.54; 2.1-3.08) and the restoration with a crown (2.1; 1.67-2.63). Associations between factors and AL were stronger and accuracy higher when using fewer complex models like decision tree (F1 score: 0.9 (0.89-0.9)). (4) Conclusions: The presence of AL was higher in root-canal treated teeth, those with crowns and molars. More complex machine learning models did not outperform less-complex ones.

Modeling strength characteristics of basalt fiber reinforced concrete using multiple explainable machine learning with a graphical user interface

This study investigated the importance of applying explainable artificial intelligence (XAI) on different machine learning (ML) models developed to predict the strength characteristics of basalt-fiber reinforced concrete (BFRC). Even though ML is widely adopted in strength prediction in concrete, the black-box nature of predictions hinders the interpretation of results. Among several attempts to overcome this limitation by using explainable AI, researchers have employed only a single explanation method. In this study, we used three tree-based ML models (Decision tree, Gradient Boosting tree, and Light Gradient Boosting Machine) to predict the mechanical strength characteristics (compressive strength, flexural strength, and tensile strength) of basal fiber reinforced concrete (BFRC). For the first time, we employed two explanation methods (Shapley additive explanations (SHAP) and local interpretable model-agnostic explanations (LIME)) to provide explanations for all models. These explainable methods reveal the underlying decision-making criteria of complex machine learning models, improving the end user's trust. The comparison highlights that tree-based models obtained good accuracy in predicting strength characteristics yet, their explanations were different either by the magnitude of feature importance or the order of importance. This disagreement pushes towards complicated decision-making based on ML predictions which further stresses (1) extending XAI-based research in concrete strength predictions, and (2) involving domain experts to evaluate XAI results. The study concludes with the development of a “user-friendly computer application” which enables quick strength prediction of basalt fiber reinforced concrete (BFRC).

Explaining any black box model using real data

In recent years the use of complex machine learning has increased drastically. These complex black box models trade interpretability for accuracy. The lack of interpretability is troubling for, e.g., socially sensitive, safety-critical, or knowledge extraction applications. In this paper, we propose a new explanation method, SLISE, for interpreting predictions from black box models. SLISE can be used with any black box model (model-agnostic), does not require any modifications to the black box model (post-hoc), and explains individual predictions (local). We evaluate our method using real-world datasets and compare it against other model-agnostic, local explanation methods. Our approach solves shortcomings in other related explanation methods by only using existing data instead of sampling new, artificial data. The method also generates more generalizable explanations and is usable without modification across various data domains.

Leakage and the reproducibility crisis in machine-learning-based science

Machine-learning (ML) methods have gained prominence in the quantitative sciences. However, there aremany known methodological pitfalls, including data leakage, in ML-based science. We systematically investigate reproducibility issues in ML-based science. Through a survey of literature in fields that haveadopted ML methods, we find 17 fields where leakage has been found, collectively affecting 294 papersand, in some cases, leading to wildly overoptimistic conclusions. Based on our survey, we introduce a detailed taxonomy of eight types of leakage, ranging from textbook errors to open research problems. We propose that researchers test for each type of leakage by filling out model info sheets, which weintroduce. Finally, we conduct a reproducibility study of civil war prediction, where complex ML models are believed to vastly outperform traditional statistical models such as logistic regression (LR). When theerrors are corrected, complex ML models do not perform substantively better than decades-old LR models.

Interpolating Nonadiabatic Molecular Dynamics Hamiltonian with Bidirectional Long Short-Term Memory Networks.

Essential for understanding far-from-equilibrium processes, nonadiabatic (NA) molecular dynamics (MD) requires expensive calculations of the excitation energies and NA couplings. Machine learning (ML) can simplify computation; however, the NA Hamiltonian requires complex ML models due to its intricate relationship to atomic geometry. Working directly in the time domain, we employ bidirectional long short-term memory networks (Bi-LSTM) to interpolate the Hamiltonian. Applying this multiscale approach to three metal-halide perovskite systems, we achieve two orders of magnitude computational savings compared to direct ab initio calculation. Reasonable charge trapping and recombination times are obtained with NA Hamiltonian sampling every half a picosecond. The Bi-LSTM-NAMD method outperforms earlier models and captures both slow and fast time scales. In combination with ML force fields, the methodology extends NAMD simulation times from picoseconds to nanoseconds, comparable to charge carrier lifetimes in many materials. Nanosecond sampling is particularly important in systems containing defects, boundaries, interfaces, etc. that can undergo slow rearrangements.

Explainable Image Classification: The Journey So Far and the Road Ahead

Explainable Artificial Intelligence (XAI) has emerged as a crucial research area to address the interpretability challenges posed by complex machine learning models. In this survey paper, we provide a comprehensive analysis of existing approaches in the field of XAI, focusing on the tradeoff between model accuracy and interpretability. Motivated by the need to address this tradeoff, we conduct an extensive review of the literature, presenting a multi-view taxonomy that offers a new perspective on XAI methodologies. We analyze various sub-categories of XAI methods, considering their strengths, weaknesses, and practical challenges. Moreover, we explore causal relationships in model explanations and discuss approaches dedicated to explaining cross-domain classifiers. The latter is particularly important in scenarios where training and test data are sampled from different distributions. Drawing insights from our analysis, we propose future research directions, including exploring explainable allied learning paradigms, developing evaluation metrics for both traditionally trained and allied learning-based classifiers, and applying neural architectural search techniques to minimize the accuracy–interpretability tradeoff. This survey paper provides a comprehensive overview of the state-of-the-art in XAI, serving as a valuable resource for researchers and practitioners interested in understanding and advancing the field.

User-Centric Explainability in Healthcare: A Knowledge-Level Perspective of Informed Machine Learning

Explaining increasingly complex machine learning will remain crucial to cope with risks, regulations, responsibilities, and human support in healthcare. However, extant explainable systems mostly provide explanations that mismatch clinical users' conceptions and fail their expectations to leverage validated and clinically relevant information. A key to more user-centric and satisfying explanations can be seen in combining data-driven and knowledge-based systems, i.e., to utilize prior knowledge jointly with the patterns learned from data. We conduct a structured review of knowledge-informed machine learning in healthcare. In this conceptual study, we build on a framework to characterize user knowledge and prior knowledge embodied in explanations. Specifically, we explicate the types and contexts of knowledge to examine the fit between knowledge-informed approaches and users. Our results highlight that knowledge-informed machine learning is a promising paradigm toI enrich former data-driven systems, yielding explanations that can increase formal understanding, convey useful medical knowledge, and are more intuitive. Although complying with medical conception, it still needs to be investigated whether knowledge-informed explanations increase medical user acceptance and trust in clinical machine learning-based information systems.