The Effect of Noise Level on the Accuracy of Causal Discovery Methods with Additive Noise Models

in Communications in Computer and Information Science (2022), 1530

In recent years a lot of research was conducted within the area of causal inference and causal learning. Many methods were developed to identify the cause-effect pairs. These methods also proved their ... [more ▼]

In recent years a lot of research was conducted within the area of causal inference and causal learning. Many methods were developed to identify the cause-effect pairs. These methods also proved their ability to successfully determine the direction of causal relationships from observational real-world data. Yet in bivariate situations, causal discovery problems remain challenging. A class of methods, that also allows tackling the bivariate case, is based on Additive Noise Models (ANMs). Unfortunately, one aspect of these methods has not received much attention until now: what is the impact of different noise levels on the ability of these methods to identify the direction of the causal relationship? This work aims to bridge this gap with the help of an empirical study. We consider a bivariate case and two specific methods Regression with Subsequent Independence Test and Identification using Conditional Variances. We perform a set of experiments with an exhaustive range of ANMs where the additive noises’ levels gradually change from 1% to 10000% of the causes’ noise level (the latter remains fixed). Additionally, we consider several different types of distributions as well as linear and non-linear ANMs. The results of the experiments show that these causal discovery methods can fail to capture the true causal direction for some levels of noise. [less ▲]

How Nonconformity Functions and Difficulty of Datasets Impact the Efficiency of Conformal Classifiers

Poster (2021, July)

Impact of model-agnostic nonconformity functions on efficiency of conformal classifiers: an extensive study

in Proceedings of Machine Learning Research (2021), 152

BacAnalytics: A Tool to Support Secondary School Examination in France

in 25th International Symposium on Intelligent Systems (ISMIS 2020) (2020, May)

Students who failed the final examination in the secondary school in France (known as baccalauréat or baccalaureate) can improve their scores by passing a remedial test. This test consists of two oral ... [more ▼]

Students who failed the final examination in the secondary school in France (known as baccalauréat or baccalaureate) can improve their scores by passing a remedial test. This test consists of two oral examinations in two subjects of the student's choice. Students announce their choice on the day of the remedial test. Additionally, the secondary education system in France is quite complex. There exist several types of baccalaureate consisting of various streams. Depending upon the stream students belong to, they have different subjects allowed to be taken during the remedial test and different coefficients associated with each of them. In this context, it becomes difficult to estimate the number of professors of each subject required for the examination. Thereby, the general practice of remedial test organization is to mobilize a large number of professors. In this paper, we present BacAnalytics - a tool that was developed to assist the rectorate of secondary schools with the organization of remedial tests for the baccalaureate. Given profiles of students and their choices of subjects for previous years, this tool builds a predictive model and estimates the number of required professors for the current year. In the paper, we present the architecture of the tool, analyze its performance, and describe its usage by the rectorate of the Academy of Nancy-Metz in Grand Est region of France in the years 2018 and 2019. BacAnalytics achieves almost 100% of prediction accuracy with approximately 25% of redundancy and was awarded a French national prize Impulsions 2018. [less ▲]

Contrast classification rules for mining local differences in medical data

in Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications (IDAACS), 2017 9th IEEE International Conference on (2017)

Sets of Contrasting Rules to Identify Trigger Factors.

in ECAI (2016)

Sets of Contrasting Rules: A Supervised Descriptive Rule Induction Pattern for Identification of Trigger Factors

in Tools with Artificial Intelligence (ICTAI), 2016 IEEE 28th International Conference on (2016)

Search for user-related features in matrix factorization-based recommender systems

in European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML/PKDD 2014), PhD Session Proceedings (2014)

Using association rules for searching levers of influence in census data

in Procedia Social and Behavioral Sciences (2013), 73

Group methods of data processing

Book published by Lulu. com (2010)