Predictive Business Process Monitoring via Generative Adversarial Nets: The Case of Next Event Prediction

Taymouri, Farbod; Rosa, Marcello La; Erfani, Sarah M.; Bozorgi, Zahra Dasht; Verenich, Ilya

doi:10.1007/978-3-030-58666-9_14

Cited by 68 publications

(96 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Both are treated separately as inputs and a form of attention layer is used to combine both hidden representations before the predictions are encoded, leading to mayor performance improvements to prior works. Taymouri et al (2020) proposed a generative adversarial architecture (GAN), where one neural network (generator) produces traces from random noise and another network (discriminator) has to differentiate between real traces and the ones created by the generator. This technique originates from the domain of computer vision, where GANs are used to create new images, that resemble real images.…”

Section: Network Architecturementioning

confidence: 99%

See 1 more Smart Citation

A systematic literature review on state-of-the-art deep learning methods for process prediction

2021

View full text Add to dashboard Cite

Process mining enables the reconstruction and evaluation of business processes based on digital traces in IT systems. An increasingly important technique in this context is process prediction. Given a sequence of events of an ongoing trace, process prediction allows forecasting upcoming events or performance measurements. In recent years, multiple process prediction approaches have been proposed, applying different data processing schemes and prediction algorithms. This study focuses on deep learning algorithms since they seem to outperform their machine learning alternatives consistently. Whilst having a common learning algorithm, they use different data preprocessing techniques, implement a variety of network topologies and focus on various goals such as outcome prediction, time prediction or control-flow prediction. Additionally, the set of log-data, evaluation metrics and baselines used by the authors diverge, making the results hard to compare. This paper attempts to synthesise the advantages and disadvantages of the procedural decisions in these approaches by conducting a systematic literature review.

show abstract

Section: Network Architecturementioning

confidence: 99%

“…For the prediction of events having a shorter prefix, the feature matrix is padded with zeros. Taymouri et al (2020) on the other hand dropped samples, when the prefix was to short. This leads to incomparable results.…”

Section: Data Preprocessing and Feature Engineeringmentioning

confidence: 99%

A systematic literature review on state-of-the-art deep learning methods for process prediction

2021

View full text Add to dashboard Cite

show abstract

“…RNN's have earlier been used in predictive process monitoring [11,22]. Recent work in predictive monitoring has also introduced the use of Generative Adverserial Nets [23]: a self-supervising machine learning technique based on training a discriminator and a generator simultaneously, which draws some similarities to the technique introduced here.…”

Section: Related Workmentioning

confidence: 99%

Supervised Conformance Checking Using Recurrent Neural Network Classifiers

Peeperkorn

Broucke

Weerdt

2021

Lecture Notes in Business Information Processing

View full text Add to dashboard Cite

Conformance checking is concerned with the task of assessing the quality of process models describing actual behavior captured in an event log across different dimensions. In this paper, a novel approach for obtaining the degree of recall and precision between a process model and event log is introduced. The approach relies on the generation of a so-called "antilog", randomly constructed from the activity vocabulary, on one hand, and a simulated "model log", which is played-out from the given model. In the case of recall the antilog and model log are used to train a recurrent neural network classifier. This network allows for calculating the probability of a trace being part of the model log or the antilog. If thereupon the event log is fed to the neural network, a value for recall can be obtained. In the case of precision the neural network is trained using a given event log and the antilog, and the model log is fed to it afterwards. We show that this new method can be used to measure global recall and precision correctly in some common examples.

show abstract

“…In lieu of RNNs, Pasquadibisceglie et al investigated Convolutional Neural Networks for the same purpose [10]. Building on top of all these approaches, Taymouri et al tackled the problem by implementing a Generative Adversarial Network, with promising results [11].…”

Section: Related Workmentioning

confidence: 99%

“…RNNs are at the core of significant progress in other fields of Computer Science such as Natural Language Processing, or Bioinformatics [5]. The PM community has recently shown significant interest in RNNs, but principally on the topic of Predictive Business Process Monitoring [6,7,8,9,10,11].…”

Section: Introductionmentioning

confidence: 99%

An Alignment Cost-Based Classification of Log Traces Using Machine-Learning

Boltenhagen

Chetioui

Huber

2021

Lecture Notes in Business Information Processing

View full text Add to dashboard Cite

Conformance checking is an important aspect of process mining that identifies the differences between the behaviors recorded in a log and those exhibited by an associated process model. Machine learning and deep learning methods perform extremely well in sequence analysis. We successfully apply both a Recurrent Neural Network and a Random Forest classifiers to the problem of evaluating whether the alignment cost of a log trace to a process model is below an arbitrary threshold, and provide a lower bound for the fitness of the process model based on the classification.

show abstract

Predictive Business Process Monitoring via Generative Adversarial Nets: The Case of Next Event Prediction

Cited by 68 publications

References 12 publications

A systematic literature review on state-of-the-art deep learning methods for process prediction

A systematic literature review on state-of-the-art deep learning methods for process prediction

Supervised Conformance Checking Using Recurrent Neural Network Classifiers

An Alignment Cost-Based Classification of Log Traces Using Machine-Learning

Contact Info

Product

Resources

About