Showing 1–7 of 7 results for author: Mijatovic, N

Unsupervised Pairwise Causal Discovery on Heterogeneous Data using Mutual Information Measures

Authors: Alexandre Trilla, Nenad Mijatovic

Abstract: A fundamental task in science is to determine the underlying causal relations because it is the knowledge of this functional structure what leads to the correct interpretation of an effect given the apparent associations in the observed data. In this sense, Causal Discovery is a technique that tackles this challenge by analyzing the statistical properties of the constituent variables. In this work… ▽ More A fundamental task in science is to determine the underlying causal relations because it is the knowledge of this functional structure what leads to the correct interpretation of an effect given the apparent associations in the observed data. In this sense, Causal Discovery is a technique that tackles this challenge by analyzing the statistical properties of the constituent variables. In this work, we target the generalizability of the discovery method by following a reductionist approach that only involves two variables, i.e., the pairwise or bi-variate setting. We question the current (possibly misleading) baseline results on the basis that they were obtained through supervised learning, which is arguably contrary to this genuinely exploratory endeavor. In consequence, we approach this problem in an unsupervised way, using robust Mutual Information measures, and observing the impact of the different variable types, which is oftentimes ignored in the design of solutions. Thus, we provide a novel set of standard unbiased results that can serve as a reference to guide future discovery tasks in completely unknown environments. △ Less

Submitted 1 August, 2024; originally announced August 2024.

Comments: 26th International Conference of the Catalan Association for Artificial Intelligence

Industrial-Grade Smart Troubleshooting through Causal Technical Language Processing: a Proof of Concept

Authors: Alexandre Trilla, Ossee Yiboe, Nenad Mijatovic, Jordi Vitrià

Abstract: This paper describes the development of a causal diagnosis approach for troubleshooting an industrial environment on the basis of the technical language expressed in Return on Experience records. The proposed method leverages the vectorized linguistic knowledge contained in the distributed representation of a Large Language Model, and the causal associations entailed by the embedded failure modes… ▽ More This paper describes the development of a causal diagnosis approach for troubleshooting an industrial environment on the basis of the technical language expressed in Return on Experience records. The proposed method leverages the vectorized linguistic knowledge contained in the distributed representation of a Large Language Model, and the causal associations entailed by the embedded failure modes and mechanisms of the industrial assets. The paper presents the elementary but essential concepts of the solution, which is conceived as a causality-aware retrieval augmented generation system, and illustrates them experimentally on a real-world Predictive Maintenance setting. Finally, it discusses avenues of improvement for the maturity of the utilized causal technology to meet the robustness challenges of increasingly complex scenarios in the industry. △ Less

Submitted 30 July, 2024; originally announced July 2024.

Comments: 2nd Workshop on Causal Inference and Machine Learning in Practice at the KDD 2024 Conference. arXiv admin note: text overlap with arXiv:2407.11056

Industrial-Grade Time-Dependent Counterfactual Root Cause Analysis through the Unanticipated Point of Incipient Failure: a Proof of Concept

Authors: Alexandre Trilla, Rajesh Rajendran, Ossee Yiboe, Quentin Possamaï, Nenad Mijatovic, Jordi Vitrià

Abstract: This paper describes the development of a counterfactual Root Cause Analysis diagnosis approach for an industrial multivariate time series environment. It drives the attention toward the Point of Incipient Failure, which is the moment in time when the anomalous behavior is first observed, and where the root cause is assumed to be found before the issue propagates. The paper presents the elementary… ▽ More This paper describes the development of a counterfactual Root Cause Analysis diagnosis approach for an industrial multivariate time series environment. It drives the attention toward the Point of Incipient Failure, which is the moment in time when the anomalous behavior is first observed, and where the root cause is assumed to be found before the issue propagates. The paper presents the elementary but essential concepts of the solution and illustrates them experimentally on a simulated setting. Finally, it discusses avenues of improvement for the maturity of the causal technology to meet the robustness challenges of increasingly complex environments in the industry. △ Less

Submitted 10 July, 2024; originally announced July 2024.

Comments: Accepted for the Causal Inference for Time Series Data Workshop at the 40th Conference on Uncertainty in Artificial Intelligence (CI4TS 2024)

A Self-Commissioning Edge Computing Method for Data-Driven Anomaly Detection in Power Electronic Systems

Authors: Pere Izquierdo Gomez, Miguel E. Lopez Gajardo, Nenad Mijatovic, Tomislav Dragicevic

Abstract: Ensuring the reliability of power electronic converters is a matter of great importance, and data-driven condition monitoring techniques are cementing themselves as an important tool for this purpose. However, translating methods that work well in controlled lab environments to field applications presents significant challenges, notably because of the limited diversity and accuracy of the lab trai… ▽ More Ensuring the reliability of power electronic converters is a matter of great importance, and data-driven condition monitoring techniques are cementing themselves as an important tool for this purpose. However, translating methods that work well in controlled lab environments to field applications presents significant challenges, notably because of the limited diversity and accuracy of the lab training data. By enabling the use of field data, online machine learning can be a powerful tool to overcome this problem, but it introduces additional challenges in ensuring the stability and predictability of the training processes. This work presents an edge computing method that mitigates these shortcomings with minimal additional memory usage, by employing an autonomous algorithm that prioritizes the storage of training samples with larger prediction errors. The method is demonstrated on the use case of a self-commissioning condition monitoring system, in the form of a thermal anomaly detection scheme for a variable frequency motor drive, where the algorithm self-learned to distinguish normal and anomalous operation with minimal prior knowledge. The obtained results, based on experimental data, show a significant improvement in prediction accuracy and training speed, when compared to equivalent models trained online without the proposed data selection process. △ Less

Submitted 5 December, 2023; originally announced December 2023.

Estimation of hysteretic losses in the HTS coils made of coated conductor tapes of an electric generator during transient operation

Authors: Víctor M. R. Zermeño, Asger B. Abrahamsen, Nenad Mijatovic, Bogi B. Jensen, Mads P. Sørensen

Abstract: In this work we present a modeling tool designed to estimate the hysteretic losses in the coils of an electric generator with coils made of coated conductor tapes during transient operation. The model is based on a two-stage segregated model approach that allows simulating the electric generator and the current distribution in the superconducting coils using a one-way coupling from the generator t… ▽ More In this work we present a modeling tool designed to estimate the hysteretic losses in the coils of an electric generator with coils made of coated conductor tapes during transient operation. The model is based on a two-stage segregated model approach that allows simulating the electric generator and the current distribution in the superconducting coils using a one-way coupling from the generator to the HTS coils model. The model has two inputs: the rotational speed and the electric load signal. A homogeneous anisotropic bulk model for the coils allows computing the current distribution in the coils. From this distribution, the hysteretic losses are estimated. Beyond the interest on providing an estimate on the global energy dissipation in the machine, in this work we present a more detailed local analysis that allows addressing issues such as coil design, critical current ratting, electric load change rate limits, cryocooler design, identification of quench-prone regions and overall transient performance. △ Less

Submitted 22 January, 2016; originally announced January 2016.

Parallel Hierarchical Affinity Propagation with MapReduce

Authors: Dillon Mark Rose, Jean Michel Rouly, Rana Haber, Nenad Mijatovic, Adrian M. Peter

Abstract: The accelerated evolution and explosion of the Internet and social media is generating voluminous quantities of data (on zettabyte scales). Paramount amongst the desires to manipulate and extract actionable intelligence from vast big data volumes is the need for scalable, performance-conscious analytics algorithms. To directly address this need, we propose a novel MapReduce implementation of the e… ▽ More The accelerated evolution and explosion of the Internet and social media is generating voluminous quantities of data (on zettabyte scales). Paramount amongst the desires to manipulate and extract actionable intelligence from vast big data volumes is the need for scalable, performance-conscious analytics algorithms. To directly address this need, we propose a novel MapReduce implementation of the exemplar-based clustering algorithm known as Affinity Propagation. Our parallelization strategy extends to the multilevel Hierarchical Affinity Propagation algorithm and enables tiered aggregation of unstructured data with minimal free parameters, in principle requiring only a similarity measure between data points. We detail the linear run-time complexity of our approach, overcoming the limiting quadratic complexity of the original algorithm. Experimental validation of our clustering methodology on a variety of synthetic and real data sets (e.g. images and point data) demonstrates our competitiveness against other state-of-the-art MapReduce clustering techniques. △ Less

Submitted 28 March, 2014; originally announced March 2014.

Calculation of AC losses in stacks and coils made of second generation high temperature superconducting tapes for large scale applications

Authors: Victor M. R. Zermeno, Asger B. Abrahamsen, Nenad Mijatovic, Bogi B. Jensen, Mads P. Soerensen

Abstract: A homogenization method to model a stack of second generation (2G) High Temperature Superconducting (HTS) tapes under AC applied transport current or magnetic field has been obtained. The idea is to find an anisotropic bulk equivalent for the stack, such that the geometrical layout of the internal alternating structures of insulating, metallic, superconducting and substrate layers is "washed" out… ▽ More A homogenization method to model a stack of second generation (2G) High Temperature Superconducting (HTS) tapes under AC applied transport current or magnetic field has been obtained. The idea is to find an anisotropic bulk equivalent for the stack, such that the geometrical layout of the internal alternating structures of insulating, metallic, superconducting and substrate layers is "washed" out while keeping the overall electromagnetic behavior of the original stack. We disregard assumptions upon the shape of the critical region and use a power law E-J relationship allowing for overcritical current densities to be considered. The method presented here allows for a computational speedup factor of up to 2 orders of magnitude when compared to full 2-D simulations taking into account the actual dimensions of the stacks without compromising accuracy. △ Less

Submitted 12 August, 2013; originally announced August 2013.