Oleg Smirnov

Filip Cornell, Oleg Smirnov, Gabriela Zarzar Gandler et al.

Dynamic graph datasets often exhibit strong temporal patterns, such as recency, which prioritizes recent interactions, and popularity, which favors frequently occurring nodes. We demonstrate that simple heuristics leveraging only these patterns can perform on par or outperform state-of-the-art neural network models under standard evaluation protocols. To further explore these dynamics, we introduce metrics that quantify the impact of recency and popularity across datasets. Our experiments on BenchTemp and the Temporal Graph Benchmark show that our approaches achieve state-of-the-art performance across all datasets in the latter and secure top ranks on multiple datasets in the former. These results emphasize the importance of refined evaluation schemes to enable fair comparisons and promote the development of more robust temporal graph models. Additionally, they reveal that current deep learning methods often struggle to capture the key patterns underlying predictions in real-world temporal graphs. For reproducibility, we have made our code publicly available.

Sofiane Ennadir, Tianze Wang, Oleg Smirnov et al.

Normalization and scaling are fundamental preprocessing steps in time series modeling, yet their role in Transformer-based models remains underexplored from a theoretical perspective. In this work, we present the first formal analysis of how different normalization strategies, specifically instance-based and global scaling, impact the expressivity of Transformer-based architectures for time series representation learning. We propose a novel expressivity framework tailored to time series, which quantifies a model's ability to distinguish between similar and dissimilar inputs in the representation space. Using this framework, we derive theoretical bounds for two widely used normalization methods: Standard and Min-Max scaling. Our analysis reveals that the choice of normalization strategy can significantly influence the model's representational capacity, depending on the task and data characteristics. We complement our theory with empirical validation on classification and forecasting benchmarks using multiple Transformer-based models. Our results show that no single normalization method consistently outperforms others, and in some cases, omitting normalization entirely leads to superior performance. These findings highlight the critical role of preprocessing in time series learning and motivate the need for more principled normalization strategies tailored to specific tasks and datasets.

Sofiane Ennadir, Oleg Smirnov, Yassine Abbahaddou et al.

Graph Neural Networks (GNNs) have achieved strong performance across a range of graph representation learning tasks, yet their adversarial robustness in graph classification remains underexplored compared to node classification. While most existing defenses focus on the message-passing component, this work investigates the overlooked role of pooling operations in shaping robustness. We present a theoretical analysis of standard flat pooling methods (sum, average and max), deriving upper bounds on their adversarial risk and identifying their vulnerabilities under different attack scenarios and graph structures. Motivated by these insights, we propose \textit{Robust Singular Pooling (RS-Pool)}, a novel pooling strategy that leverages the dominant singular vector of the node embedding matrix to construct a robust graph-level representation. We theoretically investigate the robustness of RS-Pool and interpret the resulting bound leading to improved understanding of our proposed pooling operator. While our analysis centers on Graph Convolutional Networks (GCNs), RS-Pool is model-agnostic and can be implemented efficiently via power iteration. Empirical results on real-world benchmarks show that RS-Pool provides better robustness than the considered pooling methods when subject to state-of-the-art adversarial attacks while maintaining competitive clean accuracy. Our code is publicly available at:\href{https://github.com/king/rs-pool}{https://github.com/king/rs-pool}.

Oleg Smirnov, Ameya Ketkar, Timofey Bryksin et al.

Developers often change types of program elements. Such refactoring often involves updating not only the type of the element itself, but also the API of all type-dependent references in the code, thus it is tedious and time-consuming. Despite type changes being more frequent than renamings, just a few current IDE tools provide partially-automated support only for a small set of hard-coded types. Researchers have recently proposed a data-driven approach to inferring API rewrite rules for type change patterns in Java using code commits history. In this paper, we build upon these recent advances and introduce IntelliTC - a tool to perform Java type change refactoring. We implemented it as a plugin for IntelliJ IDEA, a popular Java IDE developed by JetBrains. We present 3 different ways of providing support for such a refactoring from the standpoint of the user experience: Classic mode, Suggested Refactoring, and Inspection mode. To evaluate these modalities of using IntelliTC, we surveyed 22 experienced software developers. They positively rated the usefulness of the tool. The source code and distribution of the plugin are available on GitHub: https://github.com/JetBrains-Research/data-driven-type-migration. A demonstration video is on YouTube: https://youtu.be/pdcfvADA1PY.

Oleg Smirnov, Artyom Lobanov, Yaroslav Golubev et al.

Many code changes that developers make in their projects are repeated and constitute recurrent change patterns. It is of interest to collect such patterns from the version history of open-source repositories and suggest the most useful of them as quick fixes. In this paper, we present Revizor - a tool aimed to build custom plugins for PyCharm, a popular Python IDE. A Revizor-based plugin can take change patterns and highlight potential places for their application in the developer's code editor. If the developer accepts the quick fix, the plugin automatically performs the edit. Our approach uses a graph-based representation of code changes, which allows it to support complex distributed code patterns. Experienced developers have also rated the usability and the performance of such Revizor-based plugin positively. The source code of the tool and test plugin prototype are available on GitHub: https://github.com/JetBrains-Research/revizor. A demonstration video with a short tool description can be found on YouTube: https://youtu.be/5eLs14nco7E.

Oleg Smirnov

This paper presents tensorflow-riemopt, a Python library for geometric machine learning in TensorFlow. The library provides efficient implementations of neural network layers with manifold-constrained parameters, geometric operations on Riemannian manifolds, and stochastic optimization algorithms for non-Euclidean spaces. Designed for integration with TensorFlow Extended, it supports both research prototyping and production deployment of machine learning pipelines. The code and documentation are distributed under the MIT license and available at https://github.com/master/tensorflow-riemopt

Tianze Wang, Sofiane Ennadir, John Pertoft et al.

Time series foundation models (TSFMs) have shown strong results on public benchmarks, prompting comparisons to a "BERT moment" for time series. Their effectiveness in industrial settings, however, remains uncertain. We examine why TSFMs often struggle to generalize and highlight spectral shift (a mismatch between the dominant frequency components in downstream tasks and those represented during pretraining) as a key factor. We present evidence from an industrial-scale player engagement prediction task in mobile gaming, where TSFMs underperform domain-adapted baselines. To isolate the mechanism, we design controlled synthetic experiments contrasting signals with seen versus unseen frequency bands, observing systematic degradation under spectral mismatch. These findings position frequency awareness as critical for robust TSFM deployment and motivate new pretraining and evaluation protocols that explicitly account for spectral diversity.

Oleg Smirnov, Aditya Tewari

In this paper, we introduce a new dataset of room interior pictures with overlaying and scene text, totalling to 4836 annotated images in 25 product categories. We provide details on the collection and annotation process of our dataset, and analyze its statistics. Furthermore, we propose a baseline method for overlaying text detection, that leverages the character region-aware text detection framework to guide the classification model. We validate our approach and show its efficiency in terms of binary classification metrics, reaching the final performance of 0.95 F1 score, with false positive and false negative rates of 0.02 and 0.06 correspondingly.

Tianze Wang, Maryam Honari-Jahromi, Styliani Katsarou et al.

Methods for learning latent user representations from historical behavior logs have gained traction for recommendation tasks in e-commerce, content streaming, and other settings. However, this area still remains relatively underexplored in video and mobile gaming contexts. In this work, we present a novel method for overcoming this limitation by extending a long-range Transformer model from the natural language processing domain to player behavior data. We discuss specifics of behavior tracking in games and propose preprocessing and tokenization approaches by viewing in-game events in an analogous way to words in sentences, thus enabling learning player representations in a self-supervised manner in the absence of ground-truth annotations. We experimentally demonstrate the efficacy of the proposed approach in fitting the distribution of behavior events by evaluating intrinsic language modeling metrics. Furthermore, we qualitatively analyze the emerging structure of the learned embedding space and show its value for generating insights into behavior patterns to inform downstream applications.

Sofiane Ennadir, Gabriela Zarzar Gandler, Filip Cornell et al.

Graphs are ubiquitous in real-world applications, ranging from social networks to biological systems, and have inspired the development of Graph Neural Networks (GNNs) for learning expressive representations. While most research has centered on static graphs, many real-world scenarios involve dynamic, temporally evolving graphs, motivating the need for Continuous-Time Dynamic Graph (CTDG) models. This paper provides a comprehensive review of Graph Representation Learning (GRL) on CTDGs with a focus on Self-Supervised Representation Learning (SSRL). We introduce a novel theoretical framework that analyzes the expressivity of CTDG models through an Information-Flow (IF) lens, quantifying their ability to propagate and encode temporal and structural information. Leveraging this framework, we categorize existing CTDG methods based on their suitability for different graph types and application scenarios. Within the same scope, we examine the design of SSRL methods tailored to CTDGs, such as predictive and contrastive approaches, highlighting their potential to mitigate the reliance on labeled data. Empirical evaluations on synthetic and real-world datasets validate our theoretical insights, demonstrating the strengths and limitations of various methods across long-range, bi-partite and community-based graphs. This work offers both a theoretical foundation and practical guidance for selecting and developing CTDG models, advancing the understanding of GRL in dynamic settings.

Filip Cornell, Oleg Smirnov, Gabriela Zarzar Gandler et al.

Recent work has questioned the reliability of graph learning benchmarks, citing concerns around task design, methodological rigor, and data suitability. In this extended abstract, we contribute to this discussion by focusing on evaluation strategies in Temporal Link Prediction (TLP). We observe that current evaluation protocols are often affected by one or more of the following issues: (1) inconsistent sampled metrics, (2) reliance on hard negative sampling often introduced as a means to improve robustness, and (3) metrics that implicitly assume equal base probabilities across source nodes by combining predictions. We support these claims through illustrative examples and connections to longstanding concerns in the recommender systems community. Our ongoing work aims to systematically characterize these problems and explore alternatives that can lead to more robust and interpretable evaluation. We conclude with a discussion of potential directions for improving the reliability of TLP benchmarks.

Sofiane Ennadir, Levente Zólyomi, Oleg Smirnov et al.

Transformer models have become the dominant backbone for sequence modeling, leveraging self-attention to produce contextualized token representations. These are typically aggregated into fixed-size vectors via pooling operations for downstream tasks. While much of the literature has focused on attention mechanisms, the role of pooling remains underexplored despite its critical impact on model behavior. In this paper, we introduce a theoretical framework that rigorously characterizes the expressivity of Transformer-based models equipped with widely used pooling methods by deriving closed-form bounds on their representational capacity and the ability to distinguish similar inputs. Our analysis extends to different variations of attention formulations, demonstrating that these bounds hold across diverse architectural variants. We empirically evaluate pooling strategies across tasks requiring both global and local contextual understanding, spanning three major modalities: computer vision, natural language processing, and time-series analysis. Results reveal consistent trends in how pooling choices affect accuracy, sensitivity, and optimization behavior. Our findings unify theoretical and empirical perspectives, providing practical guidance for selecting or designing pooling mechanisms suited to specific tasks. This work positions pooling as a key architectural component in Transformer models and lays the foundation for more principled model design beyond attention alone.

Finn Rietz, Oleg Smirnov, Sara Karimi et al.

Prompting has emerged as the dominant paradigm for adapting large, pre-trained transformer-based models to downstream tasks. The Prompting Decision Transformer (PDT) enables large-scale, multi-task offline Reinforcement Learning (RL) pre-training by leveraging stochastic trajectory prompts to identify the target task. However, these prompts are sampled uniformly from expert demonstrations, overlooking a critical limitation: not all prompts are equally informative for differentiating between tasks. This limits generalization and adaptation, especially in low-data or open-world settings where sample efficiency is crucial. To address this issue, we propose a lightweight, inference-time, bandit-based prompt-tuning framework. The bandit explores and optimizes trajectory prompt selection to enhance task performance, while avoiding costly fine-tuning of the transformer backbone. Our experiments indicate not only clear performance gains due to bandit-based prompt-tuning, but also better sample complexity, scalability, and prompt space exploration compared to prompt-tuning baselines. These results highlights the importance of adaptive prompt selection mechanisms for efficient generalization in offline multi-task RL.

Finn Rietz, Oleg Smirnov, Sara Karimi et al.

Prompt tuning has emerged as a key technique for adapting large pre-trained Decision Transformers (DTs) in offline Reinforcement Learning (RL), particularly in multi-task and few-shot settings. The Prompting Decision Transformer (PDT) enables task generalization via trajectory prompts sampled uniformly from expert demonstrations -- without accounting for prompt informativeness. In this work, we propose a bandit-based prompt-tuning method that learns to construct optimal trajectory prompts from demonstration data at inference time. We devise a structured bandit architecture operating in the trajectory prompt space, achieving linear rather than combinatorial scaling with prompt size. Additionally, we show that the pre-trained PDT itself can serve as a powerful feature extractor for the bandit, enabling efficient reward modeling across various environments. We theoretically establish regret bounds and demonstrate empirically that our method consistently enhances performance across a wide range of tasks, high-dimensional environments, and out-of-distribution scenarios, outperforming existing baselines in prompt tuning.

Levente Zólyomi, Tianze Wang, Sofiane Ennadir et al.

The proliferation of digital interactions across diverse domains, such as healthcare, e-commerce, gaming, and finance, has resulted in the generation of vast volumes of event stream (ES) data. ES data comprises continuous sequences of timestamped events that encapsulate detailed contextual information relevant to each domain. While ES data holds significant potential for extracting actionable insights and enhancing decision-making, its effective utilization is hindered by challenges such as the scarcity of labeled data and the fragmented nature of existing research efforts. Self-Supervised Learning (SSL) has emerged as a promising paradigm to address these challenges by enabling the extraction of meaningful representations from unlabeled ES data. In this survey, we systematically review and synthesize SSL methodologies tailored for ES modeling across multiple domains, bridging the gaps between domain-specific approaches that have traditionally operated in isolation. We present a comprehensive taxonomy of SSL techniques, encompassing both predictive and contrastive paradigms, and analyze their applicability and effectiveness within different application contexts. Furthermore, we identify critical gaps in current research and propose a future research agenda aimed at developing scalable, domain-agnostic SSL frameworks for ES modeling. By unifying disparate research efforts and highlighting cross-domain synergies, this survey aims to accelerate innovation, improve reproducibility, and expand the applicability of SSL to diverse real-world ES challenges.

Tianze Wang, Maryam Honari-Jahromi, Styliani Katsarou et al.

This pilot study explores the application of language models (LMs) to model game event sequences, treating them as a customized natural language. We investigate a popular mobile game, transforming raw event data into textual sequences and pretraining a Longformer model on this data. Our approach captures the rich and nuanced interactions within game sessions, effectively identifying meaningful player segments. The results demonstrate the potential of self-supervised LMs in enhancing game design and personalization without relying on ground-truth labels.

Oleg Smirnov, Labinot Polisi

In this paper, we introduce the Behavior Structformer, a method for modeling user behavior using structured tokenization within a Transformer-based architecture. By converting tracking events into dense tokens, this approach enhances model training efficiency and effectiveness. We demonstrate its superior performance through ablation studies and benchmarking against traditional tabular and semi-structured baselines. The results indicate that structured tokenization with sequential processing significantly improves behavior modeling.

Tongjie Wang, Yaroslav Golubev, Oleg Smirnov et al.

Similarly to production code, code smells also occur in test code, where they are called test smells. Test smells have a detrimental effect not only on test code but also on the production code that is being tested. To date, the majority of the research on test smells has been focusing on programming languages such as Java and Scala. However, there are no available automated tools to support the identification of test smells for Python, despite its rapid growth in popularity in recent years. In this paper, we strive to extend the research to Python, build a tool for detecting test smells in this language, and conduct an empirical analysis of test smells in Python projects. We started by gathering a list of test smells from existing research and selecting test smells that can be considered language-agnostic or have similar functionality in Python's standard Unittest framework. In total, we identified 17 diverse test smells. Additionally, we searched for Python-specific test smells by mining frequent code change patterns that can be considered as either fixing or introducing test smells. Based on these changes, we proposed our own test smell called Suboptimal assert. To detect all these test smells, we developed a tool called PyNose in the form of a plugin to PyCharm, a popular Python IDE. Finally, we conducted a large-scale empirical investigation aimed at analyzing the prevalence of test smells in Python code. Our results show that 98% of the projects and 84% of the test suites in the studied dataset contain at least one test smell. Our proposed Suboptimal assert smell was detected in as much as 70.6% of the projects, making it a valuable addition to the list.

Simon Perkins, Patrick Marais, Jonathan Zwart et al.

We present Montblanc, a GPU implementation of the Radio interferometer measurement equation (RIME) in support of the Bayesian inference for radio observations (BIRO) technique. BIRO uses Bayesian inference to select sky models that best match the visibilities observed by a radio interferometer. To accomplish this, BIRO evaluates the RIME multiple times, varying sky model parameters to produce multiple model visibilities. Chi-squared values computed from the model and observed visibilities are used as likelihood values to drive the Bayesian sampling process and select the best sky model. As most of the elements of the RIME and chi-squared calculation are independent of one another, they are highly amenable to parallel computation. Additionally, Montblanc caters for iterative RIME evaluation to produce multiple chi-squared values. Modified model parameters are transferred to the GPU between each iteration. We implemented Montblanc as a Python package based upon NVIDIA's CUDA architecture. As such, it is easy to extend and implement different pipelines. At present, Montblanc supports point and Gaussian morphologies, but is designed for easy addition of new source profiles. Montblanc's RIME implementation is performant: On an NVIDIA K40, it is approximately 250 times faster than MeqTrees on a dual hexacore Intel E5-2620v2 CPU. Compared to the OSKAR simulator's GPU-implemented RIME components it is 7.7 and 12 times faster on the same K40 for single and double-precision floating point respectively. However, OSKAR's RIME implementation is more general than Montblanc's BIRO-tailored RIME. Theoretical analysis of Montblanc's dominant CUDA kernel suggests that it is memory bound. In practice, profiling shows that is balanced between compute and memory, as much of the data required by the problem is retained in L1 and L2 cache.