Ivan Zolin

Mr. Ivan Zolin

Institute of Applied Computer Science, ITMO University, St Petersburg, Russia
orcid

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
Verified
Journal Article
Open access
PDF file
VSVladimir Sergeevich Skvortsov
Vladimir Sergeevich Skvortsov
Ivan ZolinIvan Zolin
Ivan Zolin
Viacheslav Sergeevich ChukanovViacheslav Sergeevich Chukanov
Viacheslav Sergeevich Chukanov
EPEkaterina Pchitskaya
Ekaterina Pchitskaya
In biomedicine, answers must be complete, current and verifiable, yet a general-purpose large language model (LLM) can give confident but wrong or outdated answers with no sources to check. We present NeuroRAG, a multi-agent retrieval-augmented system that answers neurobiology questions using retrieved scientific sources. Its knowledge base combines a hierarchical index of neurobiology textbooks with live search of PubMed, arXiv, bioRxiv/medRxiv and NCBI Gene/Protein sources, and web search as a fallback. A router selects the sources, query-rewriting agents reformulate the question, passages are fused, reranked and compressed, three LLMs draft answers, which are merged, and, as the last step, graders check grounding in the retrieved text and relevance to the question. On benchmarks, NeuroRAG had the highest accuracy on seven of nine MMLU subsets; on PubMed Summary QA, its similarity to reference answers was comparable to Llama 3.3 70B and lower than GPT-4o and GPT-4.1. Against two published RAG pipelines, it beat FlashRAG Naive RAG on six of seven metrics and was indistinguishable from MedRAG. Because such benchmarks are limited to text and age quickly, we ran a blind evaluation of the system by 14 evaluators: NeuroRAG was chosen in 64.4%, GPT-5 in 14.1%, Claude 4.5 Sonnet in 11.0%, and none in 10.4%. This shows expert preference for NeuroRAG over the other models, though not a direct measure of factual correctness.

30 September 2024
Verified
Conference Paper
Closed access
Ivan ZolinIvan Zolin
Ivan Zolin
Viacheslav Sergeevich ChukanovViacheslav Sergeevich Chukanov
Viacheslav Sergeevich Chukanov
EPEkaterina Pchitskaya
Ekaterina Pchitskaya
Fluorescence microscopy is a technique for obtaining images of luminous objects of small size. It is widely used in fields ranging from materials science to neurobiology. Fluorescence microscopy has several advantages over other forms of microscopy, offering high sensitivity and specificity. However, it often results in images with noise and distortions, complicating subsequent analysis. This paper introduces the TriDeFusion algorithm for 3D image denoising, integrating Non-Local Means (NLM) and Modified Noise2Noise (N2N) techniques. Our results show that TriDeFusion significantly improves image quality, particularly in preserving details while reducing noise. In experiments with synthetic data, the combined methods outperformed individual approaches in both Root Mean Square Error (RMSE) and Peak Signal-to-Noise Ratio (PSNR) metrics, achieving up to a 54% reduction in RMSE and a 20% increase in PSNR. For real data, our algorithm demonstrated a significant reduction of noise mean intensity by over 50% and variance by 33%, confirming its robustness and effectiveness across different noise levels and data types.