Decision Lists Research Articles

Ways to improve the effectiveness of appeal in pre-trial criminal proceedings

The article is devoted to the study of certain issues which, if resolved, will contribute to the enhancement of the effectiveness of appeal in pre-trial criminal proceedings. It is stated that regulatory gaps and the lack of uniform approaches to the grounds, conditions and procedure for applying the appeal institute do not allow for an effective mechanism of legal protection of violated substantive and procedural human rights. The author determines that formalized approaches to the appeal procedure have a negative impact on the effective achievement of the objectives (purpose) of appeal in pre-trial criminal proceedings. It is emphasized that the appeal against a notice of suspicion has regulatory shortcomings in terms of the lack of substantive and criminal procedural grounds for filing a complaint. At the same time, the author emphasizes that there are no grounds for appealing against the decision to put a person on the wanted list. Among the defects in the procedure for applying the institution of appeal, the author highlights the lack of criteria which determine the rules of jurisdiction for consideration of appeals against decisions, actions or omissions of the prosecution during the pre-trial investigation. The author examines the impact on the effectiveness of appeal of the possibility provided by procedural law to file a complaint against decisions, actions and omissions of the prosecution to the prosecutor. It is noted that within the framework of constitutional rights, it is important to provide for the possibility of appealing against certain decisions of investigating judges which affect fundamental human rights and freedoms. In particular, the author finds that the list of appealable decisions does not include decisions on determining jurisdiction and on conducting a search in urgent cases. At the same time, the author argues that it is necessary to introduce cassation appeal in pre-trial criminal proceedings, since the right to appeal in cassation is a practical implementation of the constitutional principle of criminal justice. The author also substantiates the need to regulate the procedure for pre-trial criminal proceedings during the filing and consideration of a constitutional complaint.

Permission for concentration of business entities as a component of the state competition policy

The article examines aspects of current Ukrainian legislation, including regulations of the Antimonopoly Committee of Ukraine on concentration of business entities. The author analyses and compares them, and as a result reveals the legal mechanisms of state control over concentration processes of business entities, which consist in granting permission or prohibition of concentration by the controlling authority. The cases in which concentration is subject to a thorough examination by the supervisory authority and can be carried out only with its consent, as well as cases in which such consent is not required, are considered. The article describes the procedure for business entities to apply to the Antimonopoly Committee of Ukraine with an application for a merger clearance, and also highlights the requirements that must be met by the applicant in such an application for its consideration by the supervisory authority. The article outlines the process of consideration of cases by the Antimonopoly Committee of Ukraine on granting permits to business entities for economic concentration. The rights and obligations of applicants, third parties, interested parties and the supervisory authority during the consideration of cases are described. The list of decisions made by the committee based on the results of consideration of applications, as well as the grounds for their adoption, is considered. The author clarifies the main principles of appealing against such decisions in court or by appealing to the Cabinet of Ministers of Ukraine. The author examines the criteria and types of liability for conducting concentration without obtaining a permit, as well as for providing false information to the controlling authority in the process of obtaining a permit for concentration actions. The article examines the positive and negative factors of the State’s competition policy and legal support for control over concentration processes. The author identifies specific aspects that may have a positive impact on the effective reform of Ukrainian competition law and, as a result, ensure that Ukraine pursues an effective competition policy in the field of control over economic concentrations, which will lead to the prosperity of competition as a driving force for economic and social progress.

A General Theoretical Framework for Learning Smallest Interpretable Models

We develop a general algorithmic framework that allows us to obtain fixed-parameter tractability for computing smallest symbolic models that represent given data. Our framework applies to all ML model types that admit a certain extension property. By showing this extension property for decision trees, decision sets, decision lists, and binary decision diagrams, we obtain that minimizing these fundamental model types is fixed-parameter tractable. Our framework even applies to ensembles, which combine individual models by majority decision.

Singapore tuberculosis (TB) clinical management guidelines 2024: A modified Delphi adaptation of international guidelines for drug-susceptible TB infection and pulmonary disease.

Tuberculosis (TB) remains endemic in Singapore. Singapore's clinical practice guidelines for the management of tuberculosis were first published in 2016. Since then, there have been major new advances in the clinical management of TB, ranging from diagnostics to new drugs and treatment regimens. The National TB Programme convened a multidisciplinary panel to update guidelines for the clinical management of drug-susceptible TB infection and disease in Singapore, contextualising current evidence for local practice. Following the ADAPTE framework, the panel systematically reviewed, scored and synthesised English-language national and international TB clinical guidelines published from 2016, adapting recommendations for a prioritised list of clinical decisions. For questions related to more recent advances, an additional primary literature review was conducted via a targeted search approach. A 2-round modified Delphi process was implemented to achieve consensus for each recommendation, with a final round of edits after consultation with external stakeholders. Recommendations for 25 clinical questions spanning screening, diagnosis, selection of drug regimen, monitoring and follow-up of TB infection and disease were formulated. The availability of results from recent clinical trials led to the inclusion of shorter treatment regimens for TB infection and disease, as well as consensus positions on the role of newer technologies, such as computer-aided detection-artificial intelligence products for radiological screening of TB disease, next-generation sequencing for drug-susceptibility testing, and video observation of treatment. The panel updated recommendations on the management of drug-susceptible TB infection and disease in Singapore.

Singapore tuberculosis (TB) clinical management guidelines 2024: A modified Delphi adaptation of international guidelines for drug-susceptible TB infection and pulmonary disease

Introduction: Tuberculosis (TB) remains endemic in Singapore. Singapore’s clinical practice guidelines for the management of tuberculosis were first published in 2016. Since then, there have been major new advances in the clinical management of TB, ranging from diagnostics to new drugs and treatment regimens. The National TB Programme convened a multidisciplinary panel to update guidelines for the clinical management of drug-susceptible TB infection and disease in Singapore, contextualising current evidence for local practice. Method: Following the ADAPTE framework, the panel systematically reviewed, scored and synthesised English-language national and international TB clinical guidelines published from 2016, adapting recommendations for a prioritised list of clinical decisions. For questions related to more recent advances, an additional primary literature review was conducted via a targeted search approach. A 2-round modified Delphi process was implemented to achieve consensus for each recommendation, with a final round of edits after consultation with external stakeholders. Results: Recommendations for 25 clinical questions spanning screening, diagnosis, selection of drug regimen, monitoring and follow-up of TB infection and disease were formulated. The availability of results from recent clinical trials led to the inclusion of shorter treatment regimens for TB infection and disease, as well as consensus positions on the role of newer technologies, such as computer-aided detection-artificial intelligence products for radiological screening of TB disease, next-generation sequencing for drug-susceptibility testing, and video observation of treatment. Conclusion: The panel updated recommendations on the management of drug-susceptible TB infection and disease in Singapore.

Why Do Tree Ensemble Approximators Not Outperform the Recursive-Rule eXtraction Algorithm?

Although machine learning models are widely used in critical domains, their complexity and poor interpretability remain problematic. Decision trees (DTs) and rule-based models are known for their interpretability, and numerous studies have investigated techniques for approximating tree ensembles using DTs or rule sets, even though these approximators often overlook interpretability. These methods generate three types of rule sets: DT based, unordered, and decision list based. However, very few metrics exist that can distinguish and compare these rule sets. Therefore, the present study proposes an interpretability metric to allow for comparisons of interpretability between different rule sets and investigates the interpretability of the rules generated by the tree ensemble approximators. We compare these rule sets with the Recursive-Rule eXtraction algorithm (Re-RX) with J48graft to offer insights into the interpretability gap. The results indicate that Re-RX with J48graft can handle categorical and numerical attributes separately, has simple rules, and achieves a high interpretability, even when the number of rules is large. RuleCOSI+, a state-of-the-art method, showed significantly lower results regarding interpretability, but had the smallest number of rules.

Tax Office and Internal Affairs Bodies Cooperation in Initiating Criminal Tax Law Cases

The fight against tax crimes is carried out through cooperation of tax authorities with operational units of internal affairs bodies. In cooperation with tax authorities the operational units of internal affairs bodies (OVD) generally ensure the implementation of tax authorities powers to control and supervise compliance with legislation on taxes and fees. OVD are in fact participants in tax legal relations. At the same time, the insufficient development of legislation regulating these activities leads to legal uncertainty. In this regard, the authors propose to supplement Article 9 of the Tax Code of the Russian Federation with a clause fixing the status of internal affairs bodies as participants in relations regulated by the legislation on taxes and fees. At the same time, it is proposed to introduce some provisions into Articles 89–100 of the Tax Code of the Russian Federation regulating the powers and restrictions of an internal affairs body officer as a participant in an on-site tax audit. Changes in criminal, criminal procedure and tax legislation, which entered into force in 2022–2023, had a significant impact on the way economic security and anti-corruption units cooperate with tax authorities and preliminary investigation bodies in identifying tax crimes and initiating such criminal cases. Amid the changes in Federal legislation, the existing interdepartmental regulatory framework generally remains relevant, although it needs to be changed and updated. These measures are currently either unregulated or partially regulated in disparate departmental regulations of different legal force, which creates an objective need for the development and adoption of a single regulatory document. An acceptable orm of interdepartmental regulatory act may be a trilateral agreement on cooperation between the Investigative Committee of the Russian Federation, the Ministry of Internal Affairs of Russia and the Federal Tax Service of Russia. There is also a need to amend the Tax Code of the Russian Federation. It is important to eliminate the inconsistency of the current version of paragraph 3 of Article 32 of the Tax Code of the Russian Federation, which does not contain a complete list of decisions made by tax authorities to be sent so that to make a decision on initiating a criminal case. This results in a possibility of different interpretations of the relevant duty of tax authorities.

Seven Kinds of Decisions Sports Coaches Make

This article presents typology of coaching decisions based on the kinds of cognitive processes involved, with the aim of helping coaches improve their decisions by helping them better understand the many kinds of decisions they make. The typology has seven primary types – Snap, Simulation, Rule, Metaphor, Analogy, Story, and Pros and Cons – and another special type, Meta. The authors briefly describe each decision type and list some of their strengths and pitfalls.

Improved performance of machine learning models in predicting length of stay, discharge disposition, and inpatient mortality after total knee arthroplasty using patient-specific variables

BackgroundThis study aimed to compare the performance of ten predictive models using different machine learning (ML) algorithms and compare the performance of models developed using patient-specific vs. situational variables in predicting select outcomes after primary TKA.MethodsData from 2016 to 2017 from the National Inpatient Sample were used to identify 305,577 discharges undergoing primary TKA, which were included in the training, testing, and validation of 10 ML models. 15 predictive variables consisting of 8 patient-specific and 7 situational variables were utilized to predict length of stay (LOS), discharge disposition, and mortality. Using the best performing algorithms, models trained using either 8 patient-specific and 7 situational variables were then developed and compared.ResultsFor models developed using all 15 variables, Linear Support Vector Machine (LSVM) was the most responsive model for predicting LOS. LSVM and XGT Boost Tree were equivalently most responsive for predicting discharge disposition. LSVM and XGT Boost Linear were equivalently most responsive for predicting mortality. Decision List, CHAID, and LSVM were the most reliable models for predicting LOS and discharge disposition, while XGT Boost Tree, Decision List, LSVM, and CHAID were most reliable for mortality. Models developed using the 8 patient-specific variables outperformed those developed using the 7 situational variables, with few exceptions.ConclusionThis study revealed that performance of different models varied, ranging from poor to excellent, and demonstrated that models developed using patient-specific variables were typically better predictive of quality metrics after TKA than those developed employing situational variables.Level of EvidenceIII.

Update of the operational production system and organizational structure as the basis of the innovative development of the construction enterprise

The article presents a formalized economic-mathematical model that makes it possible to estimate the effect of the joint introduction of information modeling in construction and an integrated project management system to solve the tasks of extracting from the overall effect of the enterprises' activity the share that arises as a result of the introduction of innovations.
 It was determined that the analysis of empirical data is a mandatory condition for the formation of proposals for improving the activities of economic entities, in general, and the development of direct measures within the limits of individual functional types of activity, in particular. In order to obtain up-to-date and verified information about the activities of the studied enterprises, the work pays attention to the analysis of trends in the development of domestic construction. An analysis of the destabilizing effect of factors on the economic security of these enterprises is presented using the approaches and indicators of the rating assessment of Doing Business business conditions. This approach allows detailing the general state of the business environment at the macro level and explaining the deep processes of the response of the enterprise management system to the challenges of modernity, which form a certain security environment at different levels of management of complex socio-economic systems.
 A model of flexible management of the enterprise has been developed to ensure its economic security, which is built on the basis of two-component neural network diagnostics, taking into account the needs of the main stakeholders and determining the internal and external factors influencing the enterprise's activities. The model allows management to determine the most relevant list of management decisions and take into account external and internal threats, which ensures an increase in the level of economic security of the enterprise due to sufficient flexibility of the management system and timely response to current threats.

Assessing the predictive capacity of machine learning models using patient-specific variables in determining in-hospital outcomes after THA

BackgroundMachine learning is a subset of artificial intelligence using algorithmic modeling to progressively learn and create predictive models. Clinical application of machine learning can aid physicians through identification of risk factors and implications of predicted patient outcomes. AimsThe aim of this study was to compare patient-specific and situation perioperative variables through optimized machine learning models to predict postoperative outcomes. MethodsData from 2016 to 2017 from the National Inpatient Sample was used to identify 177,442 discharges undergoing primary total hip arthroplasty, which were included in the training, testing, and validation of 10 machine learning models. 15 predictive variables consisting of 8 patient-specific and 7 situational specific variables were utilized to predict 3 outcome variables: length of stay, discharge, and mortality. The machine learning models were assessed in responsiveness via area under the curve and reliability. ResultsFor all outcomes, Linear Support Vector Machine had the highest responsiveness among all models when using all variables. When utilizing patient-specific variables only, responsiveness of the top 3 models ranged between 0.639 and 0.717 for length of stay, 0.703–0.786 for discharge disposition, and 0.887–0.952 for mortality. The top 3 models utilizing situational variables only produced responsiveness of 0.552–0.589 for length of stay, 0.543–0.574 for discharge disposition, and 0.469–0.536 for mortality. ConclusionsLinear Support Vector Machine was the most responsive machine learning model of the 10 algorithms trained, while decision list was most reliable. Responsiveness was observed to be consistently higher with patient-specific variables than situational variables, emphasizing the predictive capacity and value of patient-specific variables. The current practice in machine learning literature generally deploys a single model, it is suboptimal to develop optimized models for application into clinical practice. The limitation of other algorithms may prohibit potential more reliable and responsive models.Level of Evidence III.

Linear threshold functions in decision lists, decision trees, and depth-2 circuits

We show that polynomial-size constant-rank linear decision trees (LDTs) can be converted to polynomial-size depth-2 threshold circuits LTF∘LTF. An intermediate construct is polynomial-size decision lists that query a conjunction of a constant number of linear threshold functions (LTFs); we show that these are equivalent to polynomial-size exact linear decision lists (ELDLs) i.e. decision lists querying exact threshold functions (ELTFs).

Determination of liability in criminal proceedings: current state

The article is devoted to the investigation of the prosecutor's change of responsibility in criminal proceedings and the problems that arise when the prosecutor uses the specified institute. Attention was drawn to the increase in the number of law enforcement agencies with investigative functions, which increased the relevance of the highlighted issue. The judicial decision on the change of jurisdiction in criminal proceedings has been analyzed. The main positions of the Criminal Court of Cassation as part of the Supreme Court regarding the determination of the jurisdiction of the State Bureau of Investigation, the National Anti-Corruption Bureau of Ukraine, the Security Service of Ukraine and the National Police of Ukraine are highlighted. Attention is focused on the decision of the Joint Chamber of the Criminal Court of Cassation as part of the Supreme Court, which defines clear criteria for recognizing the pre-trial investigation as effective. In addition, the approaches proposed by scientists regarding the application of the norms of the criminal procedural law, which regulate the institution of criminal proceedings, are summarized, and the expediency of including in the list of decisions of the investigating judge, which can be challenged during the pre-trial investigation, the decision of the investigating judge on the refusal to satisfy the complaint against the prosecutor's decision on determination of liability. The range of problems that occur when the jurisdiction of a criminal proceeding is changed and the consequences of a change of jurisdiction by the prosecutor on the body of a pre-trial investigation, which, according to Article 216 of the Criminal Procedure Code of Ukraine, is not authorized to conduct a pre-trial investigation, for example, abuse by the prosecutor of his right to change jurisdiction with from one body of pre-trial investigation to another. The importance of proper substantiation by the prosecutor of the resolution on changing jurisdiction, as well as the introduction of judicial control over the legality of the prosecutor's determination of jurisdiction for a specific body of pre-trial investigation, is highlighted.

Assessment of indicators of specific humoral immune against COVID-19 in children during the distribution of a new coronavirus infection in the Irkutsk region (2020–2021)

Background. There are many aspects of the development of immunity to the SARSCoV-2 virus, that remain poorly understood, like the influence of age-related characteristics on the intensity of immunity and the course of the disease. Studies of the state of immunity are widely used, mainly in the adults. But questions of the patho- and immunogenesis in children remain unsolved. Determining the nature of seroconversion of antibodies to SARS-CoV-2 in this age group is important information for serological monitoring for targeted immunoprophylaxis of the population and forecasting the epidemic situation in the region.The aim. Evaluation of the dynamics of seroprevalence of specific antibodies to SARS-CoV-2 in children of the Irkutsk region during the pandemic of a new coronavirus infection.Materials and methods. Study was conducted among the child population of the Irkutsk region in the period June 2020 – December 2021 as part of the Rospotrebnadzor project to assess population immunity to SARS-CoV-2 in the population of the Russian Federation. The content of antibodies to SARS-CoV-2 was determined by ELISA using native commercial test systems.Results. Population immunity among the child population of the Irkutsk region was characterized by an upward trend from 7.8 % at stage 1 to 98.4 % at stage 6 of the study. IgG to SARS-CoV-2 remained in 72.8 % and formed in 25.6 % of previously seronegative children. Antibodies were detected in 66.1 % of cases and persisted for up to 10–15 months after COVID-19 infection. The proportion of asymptomatic forms of infection among seropositive volunteers was 69.5 %, which determines the high intensity of the latent epidemic process.Conclusion. The level of seroprevalence was 98.4%. The current results of serological monitoring serve as a scientific basis for adjusting the list and scope of management decisions on the organization of preventive anti-epidemic measures, including vaccination.

Tamil Speech Synthesizer App for Android: Text Processing Module Enhancement

Objectives: Designing dynamic computer systems that are effective, efficient, simple, and satisfying to use is becoming extremely important in this age of information and communication technology. Text to Speech or Speech Synthesis is one of the many methods being investigated by researchers to improve Human-Computer Interaction. The goal here is to improve the text processing component of the Tamil voice synthesizer by including a text normalizer and loan word identification that is efficient and reliable. Methods: Text normalization is conducted on unconstrained Tamil text to turn non-standard terms into common words to reduce confusing utterances during intermediate word processing. Loan/Native words in Tamil literature are detected to enhance the Tamil voice synthesizer system’s pronunciation model. Findings: During normalization, non-standard Tamil words are replaced with standard ones to reduce ambiguous utterances during interim processing. A pronunciation model is built to improve the Tamil speech synthesizer system by identifying loan words in Tamil text. A syllable classifier is presented in this study, based on a decision list approach, which can handle various types of non-stationary sounds. Novelty: We also disclose a ’loan/native word classifier’ based on multiple linear regressions that perform well even with small words of three syllables. Such sophisticated text processors are required in today’s dominating Digital, Information-Communication Technology, and Human-Computer Interaction age. Keywords: Mobile Communication Technology; HumanComputer Interaction; Speech Synthesis Affirms; Syllable Classifier; Prerecorded database

Hardness Characterisations and Size-width Lower Bounds for QBF Resolution

We provide a tight characterisation of proof size in resolution for quantified Boolean formulas (QBF) via circuit complexity. Such a characterisation was previously obtained for a hierarchy of QBF Frege systems [ 16 ], but leaving open the most important case of QBF resolution. Different from the Frege case, our characterisation uses a new version of decision lists as its circuit model, which is stronger than the CNFs the system works with. Our decision list model is well suited to compute countermodels for QBFs. Our characterisation works for both Q-Resolution and QU-Resolution. Using our characterisation, we obtain a size-width relation for QBF resolution in the spirit of the celebrated result for propositional resolution [ 4 ]. However, our result is not just a replication of the propositional relation—intriguingly ruled out for QBF in previous research [ 12 ]—but shows a different dependence between size, width, and quantifier complexity. An essential ingredient is an improved relation between the size and width of term decision lists; this may be of independent interest. We demonstrate that our new technique elegantly reproves known QBF hardness results and unifies previous lower-bound techniques in the QBF domain.

Estimation of a Machine Learning-Based Decision Rule to Reduce Hypoglycemia Among Older Adults With Type 1 Diabetes: A Post Hoc Analysis of Continuous Glucose Monitoring in the WISDM Study.

The Wireless Innovation for Seniors with Diabetes Mellitus (WISDM) study demonstrated continuous glucose monitoring (CGM) reduced hypoglycemia over 6 months among older adults with type 1 diabetes (T1D) compared with blood glucose monitoring (BGM). We explored heterogeneous treatment effects of CGM on hypoglycemia by formulating a data-driven decision rule that selects an intervention (ie, CGM vs BGM) to minimize percentage of time <70 mg/dL for each individual WISDM participant. The precision medicine analyses used data from participants with complete data (n = 194 older adults, including those who received CGM [n = 100] and BGM [n = 94] in the trial). Policy tree and decision list algorithms were fit with 14 baseline demographic, clinical, and laboratory measures. The primary outcome was CGM-measured percentage of time spent in hypoglycemic range (<70 mg/dL), and the decision rule assigned participants to a subgroup reflecting the treatment estimated to minimize this outcome across all follow-up visits. The optimal decision rule was found to be a decision list with 3 steps. The first step moved WISDM participants with baseline time-below range >1.35% and no detectable C-peptide levels to the CGM subgroup (n = 139), and the second step moved WISDM participants with a baseline time-below range of >6.45% to the CGM subgroup (n = 18). The remaining participants (n = 37) were left in the BGM subgroup. Compared with the BGM subgroup (n = 37; 19%), the group for whom CGM minimized hypoglycemia (n = 157; 81%) had more baseline hypoglycemia, a lower proportion of detectable C-peptide, higher glycemic variability, longer disease duration, and higher proportion of insulin pump use. The decision rule underscores the benefits of CGM for older adults to reduce hypoglycemia. Diagnostic CGM and laboratory markers may inform decision-making surrounding therapeutic CGM and identify older adults for whom CGM may be a critical intervention to reduce hypoglycemia.

Dual Dynamic Programming for the Mean Standard Deviation Canadian Traveller Problem

This article studies the mean standard deviation (mean-std) Canadian traveller problem (CTP). Different from the canonical CTP, which aims at minimizing the traveller's expected travel time, while considering edge breakdown probabilities, we introduce the reliability version of CTP, which tries to find a routing policy with the minimal linear combination of the travel time's mean and standard deviation. With the recent development of internet-of-things (IoT) technology, the transportation network's edges' travel-time statistics, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">i</i> . <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">e</i> ., mean and standard deviation, as well as the traversal probabilities, are available to the end users. With those information, we propose a dual dynamic programming (DDP) method, which simultaneously estimates the first-order and the second-order moments of a given decision-list (DL) policy, and thereby makes improvements towards to the optimal one through the generalized policy iteration (GPI) scheme. We construct an open source benchmark environment to evaluate the performance of different mean-std CTP solutions, and show that the DDP method outperforms state of the arts in a range of transportation networks.

On Hard and Soft Decision Decoding of BCH Codes

The binary primitive BCH codes are cyclic and are constructed by choosing a subset of the cyclotomic cosets. Which subset is chosen determines the dimension, the minimum distance and the weight distribution of the BCH code. We construct possible BCH codes and determine their coderate, true minimum distance and the non-equivalent codes. A particular choice of cyclotomic cosets gives BCH codes which are, extended by one bit, equivalent to Reed-Muller codes, which is a known result from the sixties. We show that BCH codes have possibly better parameters than Reed-Muller codes, which are related in recent publications to polar codes. We study the decoding performance of these different BCH codes using information set decoding based on minimal weight codewords of the dual code. We show that information set decoding is possible even in case of a channel without reliability information since the decoding algorithm inherently calculates reliability information. Different BCH codes of the same rate are compared and different decoding performances and complexity are observed. Some examples of hard decision decoding of BCH codes have the same decoding performance as maximum likelihood decoding. All presented decoding methods can possibly be extended to include reliability information of a Gaussian channel for soft decision decoding. We show simulation results for soft decision list information set decoding and compare the performance to other methods.

Efficient Learning of Interpretable Classification Rules

Machine learning has become omnipresent with applications in various safety-critical domains such as medical, law, and transportation. In these domains, high-stake decisions provided by machine learning necessitate researchers to design interpretable models, where the prediction is understandable to a human. In interpretable machine learning, rule-based classifiers are particularly effective in representing the decision boundary through a set of rules comprising input features. Examples of such classifiers include decision trees, decision lists, and decision sets. The interpretability of rule-based classifiers is in general related to the size of the rules, where smaller rules are considered more interpretable. To learn such a classifier, the brute-force direct approach is to consider an optimization problem that tries to learn the smallest classification rule that has close to maximum accuracy. This optimization problem is computationally intractable due to its combinatorial nature and thus, the problem is not scalable in large datasets. To this end, in this paper we study the triangular relationship among the accuracy, interpretability, and scalability of learning rule-based classifiers. The contribution of this paper is an interpretable learning framework IMLI, that is based on maximum satisfiability (MaxSAT) for synthesizing classification rules expressible in proposition logic. IMLI considers a joint objective function to optimize the accuracy and the interpretability of classification rules and learns an optimal rule by solving an appropriately designed MaxSAT query. Despite the progress of MaxSAT solving in the last decade, the straightforward MaxSAT-based solution cannot scale to practical classification datasets containing thousands to millions of samples. Therefore, we incorporate an efficient incremental learning technique inside the MaxSAT formulation by integrating mini-batch learning and iterative rule-learning. The resulting framework learns a classifier by iteratively covering the training data, wherein in each iteration, it solves a sequence of smaller MaxSAT queries corresponding to each mini-batch. In our experiments, IMLI achieves the best balance among prediction accuracy, interpretability, and scalability. For instance, IMLI attains a competitive prediction accuracy and interpretability w.r.t. existing interpretable classifiers and demonstrates impressive scalability on large datasets where both interpretable and non-interpretable classifiers fail. As an application, we deploy IMLI in learning popular interpretable classifiers such as decision lists and decision sets. The source code is available at https://github.com/meelgroup/mlic.