Showing 1–7 of 7 results for author: Kireev, I

Universal representations for financial transactional data: embracing local, global, and external contexts

Authors: Alexandra Bazarova, Maria Kovaleva, Ilya Kuleshov, Evgenia Romanenkova, Alexander Stepikin, Alexandr Yugay, Dzhambulat Mollaev, Ivan Kireev, Andrey Savchenko, Alexey Zaytsev

Abstract: Effective processing of financial transactions is essential for banking data analysis. However, in this domain, most methods focus on specialized solutions to stand-alone problems instead of constructing universal representations suitable for many problems. We present a representation learning framework that addresses diverse business challenges. We also suggest novel generative models that accoun… ▽ More Effective processing of financial transactions is essential for banking data analysis. However, in this domain, most methods focus on specialized solutions to stand-alone problems instead of constructing universal representations suitable for many problems. We present a representation learning framework that addresses diverse business challenges. We also suggest novel generative models that account for data specifics, and a way to integrate external information into a client's representation, leveraging insights from other customers' actions. Finally, we offer a benchmark, describing representation quality globally, concerning the entire transaction history; locally, reflecting the client's current state; and dynamically, capturing representation evolution over time. Our generative approach demonstrates superior performance in local tasks, with an increase in ROC-AUC of up to 14\% for the next MCC prediction task and up to 46\% for downstream tasks from existing contrastive baselines. Incorporating external information improves the scores by an additional 20\%. △ Less

Submitted 2 April, 2024; originally announced April 2024.

Continuous-time convolutions model of event sequences

Authors: Vladislav Zhuzhel, Vsevolod Grabar, Galina Boeva, Artem Zabolotnyi, Alexander Stepikin, Vladimir Zholobov, Maria Ivanova, Mikhail Orlov, Ivan Kireev, Evgeny Burnaev, Rodrigo Rivera-Castro, Alexey Zaytsev

Abstract: Massive samples of event sequences data occur in various domains, including e-commerce, healthcare, and finance. There are two main challenges regarding inference of such data: computational and methodological. The amount of available data and the length of event sequences per client are typically large, thus it requires long-term modelling. Moreover, this data is often sparse and non-uniform, mak… ▽ More Massive samples of event sequences data occur in various domains, including e-commerce, healthcare, and finance. There are two main challenges regarding inference of such data: computational and methodological. The amount of available data and the length of event sequences per client are typically large, thus it requires long-term modelling. Moreover, this data is often sparse and non-uniform, making classic approaches for time series processing inapplicable. Existing solutions include recurrent and transformer architectures in such cases. To allow continuous time, the authors introduce specific parametric intensity functions defined at each moment on top of existing models. Due to the parametric nature, these intensities represent only a limited class of event sequences. We propose the COTIC method based on a continuous convolution neural network suitable for non-uniform occurrence of events in time. In COTIC, dilations and multi-layer architecture efficiently handle dependencies between events. Furthermore, the model provides general intensity dynamics in continuous time - including self-excitement encountered in practice. The COTIC model outperforms existing approaches on majority of the considered datasets, producing embeddings for an event sequence that can be used to solve downstream tasks - e.g. predicting next event type and return time. The code of the proposed method can be found in the GitHub repository (https://github.com/VladislavZh/COTIC). △ Less

Submitted 13 February, 2023; originally announced February 2023.

Comments: 9 pages, 3 figures

Adversarial Attacks on Deep Models for Financial Transaction Records

Authors: Ivan Fursov, Matvey Morozov, Nina Kaploukhaya, Elizaveta Kovtun, Rodrigo Rivera-Castro, Gleb Gusev, Dmitry Babaev, Ivan Kireev, Alexey Zaytsev, Evgeny Burnaev

Abstract: Machine learning models using transaction records as inputs are popular among financial institutions. The most efficient models use deep-learning architectures similar to those in the NLP community, posing a challenge due to their tremendous number of parameters and limited robustness. In particular, deep-learning models are vulnerable to adversarial attacks: a little change in the input harms the… ▽ More Machine learning models using transaction records as inputs are popular among financial institutions. The most efficient models use deep-learning architectures similar to those in the NLP community, posing a challenge due to their tremendous number of parameters and limited robustness. In particular, deep-learning models are vulnerable to adversarial attacks: a little change in the input harms the model's output. In this work, we examine adversarial attacks on transaction records data and defences from these attacks. The transaction records data have a different structure than the canonical NLP or time series data, as neighbouring records are less connected than words in sentences, and each record consists of both discrete merchant code and continuous transaction amount. We consider a black-box attack scenario, where the attack doesn't know the true decision model, and pay special attention to adding transaction tokens to the end of a sequence. These limitations provide more realistic scenario, previously unexplored in NLP world. The proposed adversarial attacks and the respective defences demonstrate remarkable performance using relevant datasets from the financial industry. Our results show that a couple of generated transactions are sufficient to fool a deep-learning model. Further, we improve model robustness via adversarial training or separate adversarial examples detection. This work shows that embedding protection from adversarial attacks improves model robustness, allowing a wider adoption of deep models for transaction records in banking and finance. △ Less

Submitted 15 June, 2021; originally announced June 2021.

CoLES: Contrastive Learning for Event Sequences with Self-Supervision

Authors: Dmitrii Babaev, Ivan Kireev, Nikita Ovsov, Mariya Ivanova, Gleb Gusev, Ivan Nazarov, Alexander Tuzhilin

Abstract: We address the problem of self-supervised learning on discrete event sequences generated by real-world users. Self-supervised learning incorporates complex information from the raw data in low-dimensional fixed-length vector representations that could be easily applied in various downstream machine learning tasks. In this paper, we propose a new method "CoLES", which adapts contrastive learning, p… ▽ More We address the problem of self-supervised learning on discrete event sequences generated by real-world users. Self-supervised learning incorporates complex information from the raw data in low-dimensional fixed-length vector representations that could be easily applied in various downstream machine learning tasks. In this paper, we propose a new method "CoLES", which adapts contrastive learning, previously used for audio and computer vision domains, to the discrete event sequences domain in a self-supervised setting. We deployed CoLES embeddings based on sequences of transactions at the large European financial services company. Usage of CoLES embeddings significantly improves the performance of the pre-existing models on downstream tasks and produces significant financial gains, measured in hundreds of millions of dollars yearly. We also evaluated CoLES on several public event sequences datasets and showed that CoLES representations consistently outperform other methods on different downstream tasks. △ Less

Submitted 22 July, 2022; v1 submitted 19 February, 2020; originally announced February 2020.

Comments: SIGMOD'22

Properties of Neutral Charmed Mesons in Proton--Nucleus Interactions at 70 GeV

Authors: SVD-2 Collaboration, :, A. N. Aleev, E. N. Ardashev, A. G. Afonin, V. P. Balandin, S. G. Basiladze, S. F. Berezhnev, G. A. Bogdanova, M. Yu. Bogolyubsky, A. M. Vishnevskaya, V. Yu. Volkov, A. P. Vorobiev, A. G. Voronin, G. G. Ermakov, P. F. Ermolov, S. N. Golovnia, S. A. Gorokhov, V. F. Golovkin, N. I. Grishin, Ya. V. Grishkevich, V. N. Zapolsky, E. G. Zverev, S. A. Zotkin, D. S. Zotkin , et al. (28 additional authors not shown)

Abstract: The results of treatment of data obtained in the SERP-E-184experiment "Investigation of mechanisms of the production of charmed particles in proton-nucleus interactions at 70 GeV and their decays" by irradiating the active target of the SVD-2 facility consisting of carbon, silicon, and lead plates, are presented. After separating a signal from the two-particle decay of neutral charmed mesons and e… ▽ More The results of treatment of data obtained in the SERP-E-184experiment "Investigation of mechanisms of the production of charmed particles in proton-nucleus interactions at 70 GeV and their decays" by irradiating the active target of the SVD-2 facility consisting of carbon, silicon, and lead plates, are presented. After separating a signal from the two-particle decay of neutral charmed mesons and estimating the cross section for charm production at a threshold energy σ(cč)=7.1 \pm 2.4(stat.) \pm 1.4(syst.) \mub/nucleon, some properties of D mesons are investigated. These include the dependence of the cross section on the target mass number (its A dependence); the behavior of the differential cross sections dσ/dpt2 and dσ/dxF; and the dependence of the parameter α on the kinematical variables xF, pt2, and plab. The experimental results in question are compared with predictions obtained on the basis of the FRITIOF7.02 code. △ Less

Submitted 8 June, 2011; originally announced June 2011.

Comments: 9 pages, 9 figures,3 tables

Journal ref: Phys.Atom.Nucl.74:324-331,2011

Neutral pion number fluctuations at high multiplicity in pp-interactions at 50 GeV

Authors: SVD-2 Collaboration, :, A. G. Afonin, A. N. Aleev, E. N. Ardashev, V. V. Avdeichikov, V. P. Balandin, S. G. Basiladze, M. A. Batouritski, S. F. Berezhnev, G. A. Bogdanova, Yu. T. Borzunov, V. A. Budilov, Yu. A. Chentsov, V. F. Golovkin, S. N. Golovnya, S. A. Gorokhov, N. I. Grishin, Ya. V. Grishkevich, G. G. Ermakov, P. F. Ermolov, N. F. Furmanets, D. E. Karmanov, A. V. Karpov, G. D. Kekelidze , et al. (47 additional authors not shown)

Abstract: Neutral pion number N0 distributions for each total number of particles in event Ntot=Nch+N0 are obtained. The scaled variance of neutral pion fluctuations, omega=D/<N0>, is measured. The fluctuations increase at Ntot >22. According to quantum statistics models it may indicate for the approaching to pion condensate conditions for high pion multiplicity in the system. Neutral pion number N0 distributions for each total number of particles in event Ntot=Nch+N0 are obtained. The scaled variance of neutral pion fluctuations, omega=D/<N0>, is measured. The fluctuations increase at Ntot >22. According to quantum statistics models it may indicate for the approaching to pion condensate conditions for high pion multiplicity in the system. △ Less

Submitted 7 June, 2011; v1 submitted 19 April, 2011; originally announced April 2011.

Comments: 8 pages, 9 figures, submitted to journal Physics of Atomic Nuclei

Proton interactions with high multiplicity

Authors: A. G. Afonin, A. N. Aleev, E. N. Ardashev, V. V. Avdeichikov, V. P. Balandin, S. G. Basiladze, M. A. Batouritski, S. F. Berezhnev, G. A. Bogdanova, Yu. T. Borzunov, V. A. Budilov, Yu. A. Chentsov, V. F. Golovkin, S. N. Golovnya, S. A. Gorokhov, N. I. Grishin, Ya. V. Grishkevich, G. G. Ermakov, P. F. Ermolov, N. F. Furmanets, D. E. Karmanov, A. V. Karpov, G. D. Kekelidze, V. I. Kireev, A. A. Kiryakov , et al. (45 additional authors not shown)

Abstract: Project Thermalization (Experiment SERP-E-190 at IHEP) is aimed to study the proton - proton interactions at 50 GeV with large number of secondary particles. In this report the experimentally measured topological cross sections are presented taking into account the detector response and procession efficiency. These data are in good agreement with gluon dominance model. The comparison with other mo… ▽ More Project Thermalization (Experiment SERP-E-190 at IHEP) is aimed to study the proton - proton interactions at 50 GeV with large number of secondary particles. In this report the experimentally measured topological cross sections are presented taking into account the detector response and procession efficiency. These data are in good agreement with gluon dominance model. The comparison with other models is also made and shows no essential discrepancies. △ Less

Submitted 1 April, 2011; originally announced April 2011.

Comments: 7 pages, 7 figures, 2 tables. Preprint IHEP 2011-4