{
  "abstract": "Selective machine unlearning aims to suppress targeted information while preserving retained and general knowledge. We investigate valence-aware gradient memory within low-rank adaptation (LoRA), separating retention-supporting and forgetting-related directions. Earlier experiments used Gamma-scaled exponential moving averages rather than genuine fractional dynamics; we preserve that distinction and introduce a normalized finite-history Grünwald–Letnikov (GL) fractional-integral accumulator with algebraically decaying coefficients. The study progresses from synthetic transformer optimization and pretrained text classification to causal language model experiments, including a new controlled evaluation on TOFU Forget10 using Qwen2.5-0.5B-Instruct. A SciQ- and WikiText-regularized teacher reduces reference-answer NLL on original and paraphrased TOFU questions while exhibiting a small World Facts penalty in a separate audit. From this fixed teacher, four unlearning methods were compared for 120 updates over three random seeds. Dual GL yields the largest paraphrased forget-set NLL increase (2.169±0.331), exceeding dual EMA (1.964±0.250), but incurs greater retain-set (1.043±0.207 versus 0.914±0.147) and World Facts (1.333±0.645 versus 1.153±0.567) degradation. Exact paired sign-flip tests do not establish statistical significance (p=0.25 for GL versus EMA). The stronger suppression achieved with fractional memory is accompanied by greater disruption of retained and general factual knowledge, suggesting a shift in model behavior rather than more selective forgetting. Whether the targeted information becomes genuinely inaccessible rather than merely less likely to be reproduced remains unresolved.",
  "abstract_shown": true,
  "affiliations": [
    "Bauman Moscow State Technical University, 105005 Moscow, Russia"
  ],
  "article_type": null,
  "arxiv": null,
  "author_total": 5,
  "authors": [
    {
      "affs": [
        0
      ],
      "name": "Aleksei Borodulin",
      "orcid": "0000-0002-9648-2395"
    },
    {
      "affs": [
        0
      ],
      "name": "Andrei Gantimurov",
      "orcid": "0009-0001-4246-9742"
    },
    {
      "affs": [
        0
      ],
      "name": "Vladimir Nelyub",
      "orcid": "0000-0003-4263-2367"
    },
    {
      "affs": [
        0
      ],
      "name": "Vadim Tynchenko",
      "orcid": "0000-0002-3959-2969"
    },
    {
      "affs": [
        0
      ],
      "name": "Ivan Malashin",
      "orcid": "0009-0008-8986-402X"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/8yrv2vjtgjx0b.bib",
  "cited_by": 0,
  "doi": "10.3390/fractalfract10110746",
  "fulltext_url": null,
  "id": "8yrv2vjtgjx0b",
  "issns": [
    "2504-3110"
  ],
  "landing_url": "https://www.mdpi.com/2504-3110/10/11/746",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": null,
  "native_subjects": [],
  "openalex": null,
  "pdf_url": null,
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-26",
  "publisher": "MDPI AG",
  "ris": "https://research.draconitic.com/paper/8yrv2vjtgjx0b.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:18:32Z",
      "native_id": "10.3390/fractalfract10110746",
      "source": "crossref",
      "url": "https://www.mdpi.com/2504-3110/10/11/746"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "issn:2504-3110",
      "subfield": "analysis-pdes"
    }
  ],
  "title": "Valence-Aware Gradient Memory and Fractional Kernel Comparisons for LoRA-Based Language Model Unlearning",
  "title_text": "Valence-Aware Gradient Memory and Fractional Kernel Comparisons for LoRA-Based Language Model Unlearning",
  "type": "peer-reviewed",
  "updated_at": "2026-10-30T12:18:32Z",
  "url": "https://research.draconitic.com/paper/8yrv2vjtgjx0b",
  "venue": "Fractal and Fractional",
  "versions": []
}