arXiv subjects

Sergey Nikolenko

Publications and source records attributed to Sergey Nikolenko.

At least 19 recordsLinked to original sources

arXiv2026

New Bounds on the Competitive Ratio of Longest Queue Drop: 1.46929591 <= CR(LQD) <= 1.683652

We study the competitive ratio of Longest Queue Drop (LQD), the canonical buffer management policy for shared memory switches, for which the previously published bounds were CR(LQD) in [1.44546086, 1.6918]. We improve both ends. For the lower bound, we provide an exact-integer certificate on a new instance family, the front-loaded family, which gives $\mathrm{CR}(\mathrm{LQD}) >= 184815365566285 / 125784985866185 = 1.46929591...$, the exact ratio on a specific finite instance evaluated against the exactly optimal offline policy. The instance is evaluated under one fixed tie rule, but the bound does not depend on the tie rule: an adaptive coupling transfers the certified value to every non-clairvoyant deterministic tie rule, and to every randomized tie rule in the adaptive adversary sense. For the upper bound, we prove that CR(LQD) <= 1.683652 by replacing the per-packet endpoint relaxation in the endgame of Antoniadis et al. (2024) with the continuum envelope relaxation of the same payment expression, which we solve exactly in closed form; unlike the lower bound, this upper bound holds for every tie rule. In the process, we find a gap in the derivation of the published proof's aggregation step (their Lemma 18). We repair the aggregation with one amortized lemma and an exact finite-head analysis; the repair restores the published 1.6918, which in turn restores the weaker conference guarantee 1.707 of Antoniadis et al. (ICALP 2021), and it also supports our further improvement.

cs.DS↗

arXiv2026

First-Order Predictable but Pairwise Fragile: Local Task Adaptation in Trained Transformers

Task arithmetic, sequential fine-tuning, activation steering, and first-order random search all operate through relatively small perturbations around an already trained checkpoint, and they rely on different local approximations: individual perturbations should be first-order predictable, task updates should compose with controlled interference, useful tangent structure should be stable and possible to estimate, and weight edits should have counterparts in representation space. We measure 8 such properties with the same harness around a multitask LoRA operating point, on 9 transformers (82M-7B), with a prospectively registered property list, thresholds, and test split. We find a shared one-direction validity window up to the tested scale $10^{-2}$, but no universal radius for pairwise composition or update ordering. Along individual directions, changes of the probe loss remain first-order predictable throughout the grid: a perturbation's effect on the loss is essentially its projection onto the gradient, which is also what makes local random search work. Pairwise structure, however, proves to be far more fragile: on over a third of the measured (model, task pair) combinations, two-update order sensitivity sets in strictly inside that window; task-gradient subspaces rotate within tens of steps; additivity under our fixed activation probe fails at full task-vector scale on several models, including both held-out 7B models; and no model median passes the registered global mean-vector weight-to-steering correspondence bar. For two sequential task-gradient steps, the leading order-dependent term is the Lie bracket $H_B\textbf{g}_A-H_A\textbf{g}_B$; its normalized prediction $c(η)=ηκ+O(η^2)$ tracks the measured defect at median ratio 1.002, while the onset scale $η^\dagger\approx0.10/κ$ spans three orders of magnitude across models and task pairs.

cs.LG↗

arXiv2026

Beyond Memory Leaderboards: Evaluating Scientific Memory as Budgeted Context Restoration

Long-term memory is becoming a core component of LLM agents, but most memory benchmarks evaluate conversations or compact summaries, while research agents need to restore evidence from full scientific papers. We introduce two full-text scientific-memory benchmarks, Public AI Memory (PAIM; 81 papers, 66 questions) and Public Transformers (PTr; 252 papers, 98 questions). We evaluate eight memory/retrieval systems, including our own proposed Theoria, plus a no-retrieval baseline. Our results show that memory leaderboards are not interpretable without the full protocol: ingestion granularity, raw-text preservation, retrieval budget, retrieval modality, rubric audit, and judge choice all affect the outcome. For example, on PAIM Graphiti wins convincingly but uses 2.6M characters of retrieved context per query, and after controlling for retrieval budget the lead disappears. On PTr, for the systems where BM25 retrieval can be added cleanly, the sparse-dense hybrid is the single most significant intervention: hybrid variants of Simple RAG, Mem0, and Theoria tie for the lead within 0.03 points. Multi-judge and human side-by-side calibration show that LLM-as-a-judge rankings are consistent across frontier judges and agree with human evaluation, with an effective resolution of roughly one point on a ten-point scale. We argue that scientific memory should be evaluated as budgeted, modality-aware context restoration rather than as an unconstrained architecture leaderboard, and we release the datasets, harness, raw outputs, judgments, and scripts to reproduce our results and serve as tools for such evaluation. Our code is available at http://gitlab.com/quantellence/research/scientific-recall-bench , and the datasets are available at http://huggingface.co/datasets/quantellence/srb-data .

cs.LG↗

arXiv2026

AgentLens: Production-Assessed Trajectory Reviews for Coding Agent Evaluation

We present AgentLens, a production-assessed benchmark for interactive code agents. Most code-agent benchmarks reduce a run to a single bit -- did the task pass? -- but the people who actually use these agents experience the entire trajectory: how the agent follows instructions, uses its tools, verifies its own work, recovers from mistakes, and talks to them along the way. AgentLens evaluates that whole trajectory. It pairs formal verification, where an objective check exists, with LLM-written trajectory reviews and side-by-side comparisons, so that each run yields a readable explanation of why the score is what it is. This makes AgentLens useful for more than ranking models: we use it to diagnose model behavior, compare successive versions of our own agent, and catch product regressions in a nightly evaluation pipeline. We release the benchmark as open source at https://github.com/agent-lens/agent-lens-bench.

cs.AI↗

arXiv2026

Recoverable but Not Stationary:Local Linear Structures in Weights and Activations

Task vectors, LoRA, activation steering, and random search around pretrained weights all suggest that learned behaviour can be controlled by linear directions. We ask which linear structures actually exist and on what scale. In a synthetic multitask transformer and LoRA adapters on DistilGPT-2 / GPT-2 we find strong local low-rank task-gradient structure but reject the fixed-task-plane hypothesis: static bases miss the recovery direction, and the useful basis drifts substantially within 100 steps. However, the first recovery updates form a trajectory-prefix basis capturing 77% of the LoRA recovery displacement. We develop random search theory with a Gaussian local-linear theorem that justifies the effectiveness of random parameter search even in very high dimensions. We also study the relation between parameter perturbations and activation steering: a single gradient step produces an activation shift with 0.58 cosine to a labelled-contrast CAA steering vector, with a similar steering effect on Qwen-0.5B BoolQ statements. We validate our results with experiments on synthetic Transformers and LLMs. Our results suggest that linear structures in trained networks are not global task directions, but evolving local geometries that partially persist across parameter and activation spaces.

cs.LG↗

arXiv2026

Sparse Prefix Caching for Hybrid and Recurrent LLM Serving

Prefix caching is a key latency optimization for autoregressive LLM serving, yet existing systems assume dense per-token key/value reuse. State-space models change the structure of the problem: a recurrent layer can resume from a single stored state rather than requiring the entire token history. This asymmetry opens a new design point between no reuse and dense caching: store exact recurrent states at a sparse set of checkpoint positions and, on a cache hit, resume from the deepest stored checkpoint and recompute the remaining suffix exactly. We formalize sparse prefix caching as checkpoint placement under a distribution over overlap depths, yielding an exact O(NM) dynamic program. For use cases where requests share a non-trivial prefix (e.g. asking different questions about a single long document), we show that our method consistently improves the Pareto frontier traced by standard heuristics on real-world data. Across QuALITY and System Prompts, distribution-aware placement dominates every fixed-budget baseline on the measured layer-group Pareto frontier and matches or outperforms the strongest heuristic (block caching) while typically using substantially fewer checkpoints, with the largest gains at low checkpoint budgets where the overlap distribution is most non-uniform. The method is most relevant when many requests share a substantial but not identical prefix within a retained cache entry. It preserves exact outputs, does not change the recurrent computation itself or require new recurrent update kernels, applies to recurrent/SSM layers whose hidden state can be extracted and restored exactly, and for hybrid models can be combined with existing KV-cache compression techniques.

cs.LG↗

arXiv2026

Beyond Isolated Clients: Integrating Graph-Based Embeddings into Event Sequence Models

Large-scale digital platforms generate billions of timestamped user-item interactions (events) that are crucial for predicting user attributes in, e.g., fraud prevention and recommendations. While self-supervised learning (SSL) effectively models the temporal order of events, it typically overlooks the global structure of the user-item interaction graph. To bridge this gap, we propose three model-agnostic strategies for integrating this structural information into contrastive SSL: enriching event embeddings, aligning client representations with graph embeddings, and adding a structural pretext task. Experiments on four financial and e-commerce datasets demonstrate that our approach consistently improves the accuracy (up to a 2.3% AUC) and reveals that graph density is a key factor in selecting the optimal integration strategy.

cs.LG↗

arXiv2025

MindFuse: Towards GenAI Explainability in Marketing Strategy Co-Creation

The future of digital marketing lies in the convergence of human creativity and generative AI, where insight, strategy, and storytelling are co-authored by intelligent systems. We present MindFuse, a brave new explainable generative AI framework designed to act as a strategic partner in the marketing process. Unlike conventional LLM applications that stop at content generation, MindFuse fuses CTR-based content AI-guided co-creation with large language models to extract, interpret, and iterate on communication narratives grounded in real advertising data. MindFuse operates across the full marketing lifecycle: from distilling content pillars and customer personas from competitor campaigns to recommending in-flight optimizations based on live performance telemetry. It uses attention-based explainability to diagnose ad effectiveness and guide content iteration, while aligning messaging with strategic goals through dynamic narrative construction and storytelling. We introduce a new paradigm in GenAI for marketing, where LLMs not only generate content but reason through it, adapt campaigns in real time, and learn from audience engagement patterns. Our results, validated in agency deployments, demonstrate up to 12 times efficiency gains, setting the stage for future integration with empirical audience data (e.g., GWI, Nielsen) and full-funnel attribution modeling. MindFuse redefines AI not just as a tool, but as a collaborative agent in the creative and strategic fabric of modern marketing.

cs.MM↗

arXiv2025

Human-Annotated NER Dataset for the Kyrgyz Language

We introduce KyrgyzNER, the first manually annotated named entity recognition dataset for the Kyrgyz language. Comprising 1,499 news articles from the 24.KG news portal, the dataset contains 10,900 sentences and 39,075 entity mentions across 27 named entity classes. We show our annotation scheme, discuss the challenges encountered in the annotation process, and present the descriptive statistics. We also evaluate several named entity recognition models, including traditional sequence labeling approaches based on conditional random fields and state-of-the-art multilingual transformer-based models fine-tuned on our dataset. While all models show difficulties with rare entity categories, models such as the multilingual RoBERTa variant pretrained on a large corpus across many languages achieve a promising balance between precision and recall. These findings emphasize both the challenges and opportunities of using multilingual pretrained models for processing languages with limited resources. Although the multilingual RoBERTa model performed best, other multilingual models yielded comparable results. This suggests that future work exploring more granular annotation schemes may offer deeper insights for Kyrgyz language processing pipelines evaluation.

cs.CL↗

arXiv2025

LAMeD: LLM-generated Annotations for Memory Leak Detection

Static analysis tools are widely used to detect software bugs and vulnerabilities but often struggle with scalability and efficiency in complex codebases. Traditional approaches rely on manually crafted annotations -- labeling functions as sources or sinks -- to track data flows, e.g., ensuring that allocated memory is eventually freed, and code analysis tools such as CodeQL, Infer, or Cooddy can use function specifications, but manual annotation is laborious and error-prone, especially for large or third-party libraries. We present LAMeD (LLM-generated Annotations for Memory leak Detection), a novel approach that leverages large language models (LLMs) to automatically generate function-specific annotations. When integrated with analyzers such as Cooddy, LAMeD significantly improves memory leak detection and reduces path explosion. We also suggest directions for extending LAMeD to broader code analysis.

cs.SE↗

arXiv2025

Against Opacity: Explainable AI and Large Language Models for Effective Digital Advertising

The opaqueness of modern digital advertising, exemplified by platforms such as Meta Ads, raises concerns regarding their autonomous control over audience targeting, pricing structures, and ad relevancy assessments. Locked in their leading positions by network effects, ``Metas and Googles of the world'' attract countless advertisers who rely on intuition, with billions of dollars lost on ineffective social media ads. The platforms' algorithms use huge amounts of data unavailable to advertisers, and the algorithms themselves are opaque as well. This lack of transparency hinders the advertisers' ability to make informed decisions and necessitates efforts to promote transparency, standardize industry metrics, and strengthen regulatory frameworks. In this work, we propose novel ways to assist marketers in optimizing their advertising strategies via machine learning techniques designed to analyze and evaluate content, in particular, predict the click-through rates (CTR) of novel advertising content. Another important problem is that large volumes of data available in the competitive landscape, e.g., competitors' ads, impede the ability of marketers to derive meaningful insights. This leads to a pressing need for a novel approach that would allow us to summarize and comprehend complex data. Inspired by the success of ChatGPT in bridging the gap between large language models (LLMs) and a broader non-technical audience, we propose a novel system that facilitates marketers in data interpretation, called SODA, that merges LLMs with explainable AI, enabling better human-AI collaboration with an emphasis on the domain of digital marketing and advertising. By combining LLMs and explainability features, in particular modern text-image models, we aim to improve the synergy between human marketers and AI systems.

cs.IR↗

arXiv2024

CCT-Code: Cross-Consistency Training for Multilingual Clone Detection and Code Search

We consider the well-known and important tasks of clone detection and information retrieval for source code. The most standard setup is to search clones inside the same language code snippets. But it is also useful to find code snippets with identical behaviour in different programming languages. Nevertheless multi- and cross-lingual clone detection has been little studied in literature. We present a novel training procedure, cross-consistency training (CCT) leveraging cross-lingual similarity, that we apply to train language models on source code in various programming languages. We show that this training is effective both for encoder- and decoder-based models. The trained encoder-based CCT-LM model achieves a new state of the art on POJ-104 (monolingual C++ clone detection benchmark) with 96.73\% MAP and AdvTest (monolingual Python code search benchmark) with 47.18\% MRR. The decoder-based CCT-LM model shows comparable performance in these tasks. In addition, we formulate the multi- and cross-lingual clone detection problem and present XCD, a new benchmark dataset produced from CodeForces submissions.

cs.CL↗

arXiv2024

$\nabla^2$DFT: A Universal Quantum Chemistry Dataset of Drug-Like Molecules and a Benchmark for Neural Network Potentials

Methods of computational quantum chemistry provide accurate approximations of molecular properties crucial for computer-aided drug discovery and other areas of chemical science. However, high computational complexity limits the scalability of their applications. Neural network potentials (NNPs) are a promising alternative to quantum chemistry methods, but they require large and diverse datasets for training. This work presents a new dataset and benchmark called $\nabla^2$DFT that is based on the nablaDFT. It contains twice as much molecular structures, three times more conformations, new data types and tasks, and state-of-the-art models. The dataset includes energies, forces, 17 molecular properties, Hamiltonian and overlap matrices, and a wavefunction object. All calculations were performed at the DFT level ($ω$B97X-D/def2-SVP) for each conformation. Moreover, $\nabla^2$DFT is the first dataset that contains relaxation trajectories for a substantial number of drug-like molecules. We also introduce a novel benchmark for evaluating NNPs in molecular property prediction, Hamiltonian prediction, and conformational optimization tasks. Finally, we propose an extendable framework for training NNPs and implement 10 models within it.

physics.chem-ph↗

arXiv2024

SOMONITOR: Combining Explainable AI & Large Language Models for Marketing Analytics

Online marketing faces formidable challenges in managing and interpreting immense volumes of data necessary for competitor analysis, content research, and strategic branding. It is impossible to review hundreds to thousands of transient online content items by hand, and partial analysis often leads to suboptimal outcomes and poorly performing campaigns. We introduce an explainable AI framework SOMONITOR that aims to synergize human intuition with AI-based efficiency, helping marketers across all stages of the marketing funnel, from strategic planning to content creation and campaign execution. SOMONITOR incorporates a CTR prediction and ranking model for advertising content and uses large language models (LLMs) to process high-performing competitor content, identifying core content pillars such as target audiences, customer needs, and product features. These pillars are then organized into broader categories, including communication themes and targeted customer personas. By integrating these insights with data from the brand's own advertising campaigns, SOMONITOR constructs a narrative for addressing new customer personas and simultaneously generates detailed content briefs in the form of user stories that, as shown in the conducted case study, can be directly applied by marketing teams to streamline content production and campaign execution. The adoption of SOMONITOR in daily operations allows digital marketers to quickly parse through extensive datasets, offering actionable insights that significantly enhance campaign effectiveness and overall job satisfaction.

cs.CY↗

arXiv2024

Toolken+: Improving LLM Tool Usage with Reranking and a Reject Option

The recently proposed ToolkenGPT tool learning paradigm demonstrates promising performance but suffers from two major issues: first, it cannot benefit from tool documentation, and second, it often makes mistakes in whether to use a tool at all. We introduce Toolken+ that mitigates the first problem by reranking top $k$ tools selected by ToolkenGPT and the second problem with a special "Reject" option such that the model will generate a vocabulary token if "Reject" is ranked first. We demonstrate the effectiveness of Toolken+ on multistep numerical reasoning and tool selection tasks.

cs.CL↗

arXiv2024

Robust AI-Generated Text Detection by Restricted Embeddings

Growing amount and quality of AI-generated texts makes detecting such content more difficult. In most real-world scenarios, the domain (style and topic) of generated data and the generator model are not known in advance. In this work, we focus on the robustness of classifier-based detectors of AI-generated text, namely their ability to transfer to unseen generators or semantic domains. We investigate the geometry of the embedding space of Transformer-based text encoders and show that clearing out harmful linear subspaces helps to train a robust classifier, ignoring domain-specific spurious features. We investigate several subspace decomposition and feature selection strategies and achieve significant improvements over state of the art methods in cross-domain and cross-generator transfer. Our best approaches for head-wise and coordinate-based subspace removal increase the mean out-of-distribution (OOD) classification score by up to 9% and 14% in particular setups for RoBERTa and BERT embeddings respectively. We release our code and data: https://github.com/SilverSolver/RobustATD

cs.CL↗

arXiv2024

Neural Click Models for Recommender Systems

We develop and evaluate neural architectures to model the user behavior in recommender systems (RS) inspired by click models for Web search but going beyond standard click models. Proposed architectures include recurrent networks, Transformer-based models that alleviate the quadratic complexity of self-attention, adversarial and hierarchical architectures. Our models outperform baselines on the ContentWise and RL4RS datasets and can be used in RS simulators to model user response for RS evaluation and pretraining.

cs.IR↗

arXiv2024

AI-generated text boundary detection with RoFT

Due to the rapid development of large language models, people increasingly often encounter texts that may start as written by a human but continue as machine-generated. Detecting the boundary between human-written and machine-generated parts of such texts is a challenging problem that has not received much attention in literature. We attempt to bridge this gap and examine several ways to adapt state of the art artificial text detection classifiers to the boundary detection setting. We push all detectors to their limits, using the Real or Fake text benchmark that contains short texts on several topics and includes generations of various language models. We use this diversity to deeply examine the robustness of all detectors in cross-domain and cross-model settings to provide baselines and insights for future research. In particular, we find that perplexity-based approaches to boundary detection tend to be more robust to peculiarities of domain-specific data than supervised fine-tuning of the RoBERTa model; we also find which features of the text confuse boundary detection algorithms and negatively influence their performance in cross-domain settings.

cs.CL↗