Pith. sign in

REVIEW 2 major objections 5 minor 300 references

Reviewed by Pith at T0; open to challenge.

T0 means a machine referee read the full paper against a public rubric. The mark states how deep the mechanical check went, never who wrote it. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

Selective forgetting in multimodal foundation models can be organized by where you intervene in the pipeline, not by algorithm family, making methods comparable across vision, language, video, and audio.

2026-07-10 15:38 UTC pith:TVZG5DYP

load-bearing objection Solid system-first survey of multimodal unlearning with real tables and a repo; the taxonomy is a useful map, not a proven advance over prior algorithm-centric reviews. the 2 major comments →

arxiv 2607.07907 v1 pith:TVZG5DYP submitted 2026-07-08 cs.LG cs.AIcs.CLcs.CRcs.MM

Multimodal Unlearning Across Vision, Language, Video, and Audio: Survey of Methods, Datasets, and Benchmarks

Nobin Sarwar , Shubhashis Roy Dipta , Zheyuan Liu , Vaidehi Patil This is my paper
classification cs.LG cs.AIcs.CLcs.CRcs.MM
keywords multimodal unlearningvision-language modelsdiffusion modelsmachine unlearningsystem-first taxonomyconcept forgettingevaluation benchmarkscross-modal privacy
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Multimodal foundation models absorb sensitive, copyrighted, biased, or unsafe cross-modal associations from web-scale training, and full retraining after each deletion request is often impossible. Multimodal unlearning aims to remove specific instances or concepts while keeping the rest of the model useful. This survey argues that the right organizing principle is a system-first taxonomy: methods are grouped by intervention stage and control pathway, with a top-level split between instance-level and concept-level forgetting. That scaffold lets researchers and practitioners compare deletion strength, retention, efficiency, reversibility, and robustness across architectures and modalities. The paper also consolidates datasets, benchmarks, evaluation metrics, applications, and open problems so the field can move toward deployable, accountable unlearning.

Core claim

A unified, system-oriented view of multimodal unlearning—organized by intervention stage (data-side, training-time, architecture-constrained, training-free, decoding-time) and control pathway, with forgetting target scope as instance versus concept—enables systematic comparison across vision, language, video, and audio and clarifies the practical trade-offs that algorithm-centric taxonomies obscure.

What carries the argument

System-first taxonomy of multimodal unlearning: methods are classified by where and how they intervene in the multimodal pipeline (data path, training, architecture, weight/representation edits, or decoding/conditioning), with instance-level versus concept-level forgetting as the primary scope split.

Load-bearing premise

The claim that grouping methods by intervention stage and control pathway is a more stable and useful scaffold for cross-modal comparison and deployment than earlier algorithm-centric taxonomies is asserted by contrast, not proven by a controlled study of how people actually use the taxonomy.

What would settle it

If independent annotators cannot place new multimodal unlearning papers into the taxonomy with high agreement, or if practitioners using it do not choose methods with better measured trade-offs (deletion strength, retention, cost) than those using algorithm-centric surveys, the central organizational claim fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Method papers can be compared on shared axes—deletion strength, utility retention, efficiency, reversibility, robustness—across VLMs, diffusion models, video, and audio rather than only within one algorithm family.
  • Deployment choices can target the right intervention point: data hygiene, training edits, architecture freezes, closed-form weight or representation edits, or reversible decoding-time controls.
  • Evaluation and benchmark design should report forgetting, safety/privacy audits, retained utility, adversarial reactivation, and compute together instead of single proxy scores.
  • Open problems—certified deletion, sequential unlearning, cross-modal leakage, frontier-scale models, and unified benchmarks—become shared research targets rather than isolated modality-specific issues.
  • Governance use cases (privacy/RTBF, safety, copyright, fairness, personalization, backdoor cleanup) can be mapped to the same intervention map for accountable model updates.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If decoding-time and training-free methods remain the only practical options at foundation scale, the field may treat unlearning more as runtime policy control than as true data-deletion guarantees.
  • Cross-modal leakage implies that unlearning only the text pathway of a VLM is likely incomplete; evaluation that never probes vision or audio recovery of the same concept will systematically overstate success.
  • A living taxonomy repository is only as valuable as versioned placement rules; without public inter-annotator protocols, the scaffold could fragment as methods multiply.
  • Sequential and continual unlearning will become the real product requirement: one-shot benchmarks may not predict whether forgotten concepts reappear after fine-tuning or repeated deletions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. This survey reviews multimodal unlearning for foundation models spanning vision, language, video, and audio. It formalizes approximate retraining equivalence via an (ε, δ) criterion and a forget/retain objective (Section 2), then organizes methods by forgetting target scope (instance vs concept) and by intervention stage and control pathway: data-side, training-time, architecture-constrained, training-free, and decoding-time (Figures 1–2, Section 3). It compiles datasets and benchmarks (Tables 2–3 and Appendix Tables 4–6), evaluation metrics (Figure 3, Appendix B), applications (Figure 4, Appendix E), open challenges, and future directions, and releases a curated repository. The central claim is that this system-first taxonomy enables systematic cross-modal comparison and clarifies trade-offs among deletion strength, retention, efficiency, reversibility, and robustness relative to prior algorithm-centric surveys (Abstract, Introduction, Table 1).

Significance. If the organizational claim holds, the paper would be a useful reference for a fast-moving area that currently lacks a unified cross-modal map. Strengths include a coherent formal setup in Section 2, a clear intervention-stage taxonomy with representative citations in Section 3, careful compilation of datasets/benchmarks/metrics (Tables 2–3, Appendix B), and an open repository. These assets can help practitioners locate methods by deployment control point and help the community track gaps (especially audio/video). The contribution is primarily organizational and bibliographic rather than a new theorem, algorithm, or empirical result; its lasting value depends on whether the taxonomy is adopted as a stable scaffold.

major comments (2)
  1. Abstract, Introduction, and Table 1 claim that the system-first taxonomy “enables systematic comparison” and “clarifies trade-offs” among deletion strength, retention, efficiency, reversibility, and robustness. Figures 1–2 and §3 organize methods by intervention stage, but the manuscript does not demonstrate that methods sharing a control pathway share similar trade-off profiles, nor does it provide any controlled comparison (e.g., method-selection utility, inter-annotator agreement on placement, or stability under re-labeling) against algorithm-centric taxonomies. Without such evidence the strongest claim reduces to a well-organized literature map; either add a compact comparative analysis (even qualitative, with explicit trade-off columns per category) or soften the claim to “organizes methods by intervention stage to facilitate comparison.”
  2. The title and Abstract advertise coverage “across vision, language, video, and audio,” yet Limitations and the dataset tables show audio and video remain comparatively thin (e.g., Table 2 lists few audio/video entries; many video/audio methods appear only as brief citations in §3.2–3.4). This imbalance is acknowledged but not reflected in the strength of the cross-modal claims. Please either expand the audio/video synthesis with explicit cross-modal transfer lessons, or qualify the title/Abstract so that the primary evidence base (image–text VLMs and diffusion) is transparent.
minor comments (5)
  1. Table 1 marks “Ours ACL’26” while the arXiv header is 8 Jul 2026; clarify venue status (submitted/accepted/under review) to avoid confusion.
  2. Section 2 introduces image–text pairs then states generalization to video/audio; a short explicit note on how Df/Dr and the (ε, δ) criterion lift to temporal or multi-track modalities would help readers apply the formalization.
  3. Figure 2 is dense; a small legend or color coding distinguishing instance-level vs concept-level methods would improve readability.
  4. Appendix B metric definitions are valuable but long; a one-page summary table mapping each metric family to the trade-offs named in the Abstract would better support the “clarifies trade-offs” claim.
  5. Minor consistency: “V oigt” / “V on dem Bussche” spacing and occasional hyphenation variants (e.g., “text-to-image” vs “text to image”) should be normalized.

Circularity Check

0 steps flagged

Survey taxonomy and formalization are organizational synthesis, not a derivation that reduces to its inputs by construction.

full rationale

This is a literature survey. Its load-bearing product is a system-first taxonomy (intervention stage and control pathway; instance- vs concept-level forgetting) plus a standard formalization of multimodal unlearning (forget/retain sets, approximate retraining equivalence, two-term objective). Section 2 restates classical unlearning criteria (Cao & Yang; Bourtoule et al.; DP-style (ε,δ) bounds) and modality-specific loss templates drawn from the cited primary literature; none of these equations fit a free parameter to data and then re-label the fit as a prediction, nor define X in terms of Y while claiming to derive Y from X. Table 1 and Figures 1–2 organize external methods; they do not force trade-off conclusions by definition. Author-overlapping citations (e.g., Liu et al. prior unlearning surveys/benchmarks, Patil et al. UnLOK-VQA) appear as ordinary coverage of the field and are not used as uniqueness theorems or sole justification for the taxonomy. Asserted superiority of the scaffold over algorithm-centric taxonomies is an unvalidated usefulness claim, not circularity. No self-definitional loop, fitted-input-as-prediction, ansatz-via-self-citation, or renaming-as-derivation is present. Score 0 with empty steps is the correct outcome.

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 1 invented entities

As a survey, the paper rests on standard machine-unlearning formalisms and the organizational premise that intervention stage is the right primary axis. It introduces no fitted constants and no new physical entities; the main invented construct is the taxonomy itself as a comparison scaffold.

axioms (4)
  • domain assumption Approximate unlearning is formalized via (ε, δ) closeness of the unlearned model distribution to retraining without the forget set, mirroring differential-privacy-style stability.
    Section 2 adopts this criterion as the theoretical target for multimodal unlearning; it is standard in the unlearning literature but not proved here for multimodal foundation models.
  • domain assumption Knowledge targeted for deletion is distributed across shared multimodal representations, so selective removal must balance forget and retain objectives.
    Stated in the introduction and formalization as the reason retraining is hard and unlearning is necessary.
  • ad hoc to paper Intervention stage and control pathway form a more stable and deployment-relevant taxonomy than algorithm-centric optimization families.
    Core organizational premise of the survey (Introduction, Figures 1–2, Table 1); asserted by contrast with prior surveys rather than independently validated.
  • domain assumption Image-text formalization generalizes to video and audio for the purposes of the survey framework.
    Section 2 states the image-text setup is used for simplicity and generalizes; this is a modeling convenience, not demonstrated exhaustively.
invented entities (1)
  • System-first multimodal unlearning taxonomy (intervention stage × control pathway, with instance vs concept scope) no independent evidence
    purpose: Provide a unified scaffold for comparing methods across VLMs, diffusion models, LLMs, AFMs, video, and audio.
    The taxonomy is the paper's main organizational invention; categories group existing methods rather than introduce a new physical or computational object with independent empirical existence.

pith-pipeline@v1.1.0-grok45 · 40139 in / 2800 out tokens · 25973 ms · 2026-07-10T15:38:00.687173+00:00 · methodology

0 comments
read the original abstract

With the growing adoption of VLMs, DMs, LLMs, and AFMs, these multimodal foundation models can inadvertently encode sensitive, copyrighted, biased, or unsafe cross-modal associations that originate from their training data. Retraining after deletion requests or policy updates is often impractical, and targeted forgetting remains difficult because knowledge is distributed across shared representations. Multimodal unlearning addresses this challenge by enabling selective removal across modalities while retaining overall utility. This survey offers a unified, system-oriented view of multimodal unlearning across vision, language, audio, and video, grounded in recent advances, emerging applications, and open problems. Our taxonomy enables systematic comparison across model architectures and modalities, clarifying trade-offs among deletion strength, retention, efficiency, reversibility, and robustness. This survey highlights open problems and practical considerations to support future research and deployment of multimodal unlearning. We release a curated repository: https://smsnobin77.github.io/Awesome-Multimodal-Unlearning/

Figures

Figures reproduced from arXiv: 2607.07907 by Nobin Sarwar, Shubhashis Roy Dipta, Vaidehi Patil, Zheyuan Liu.

Figure 1
Figure 1. Figure 1: Unlearning intervention points for a Mul [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Taxonomy of multimodal unlearning methods, organized by intervention stage and control pathway, with [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Evaluation dimensions and representative met [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Key open challenges in multimodal unlearning [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

300 extracted references · 300 canonical work pages · 72 internal anchors

  1. [1]

    Gemini: A Family of Highly Capable Multimodal Models

    Gemini: a family of highly capable multimodal models , author=. arXiv preprint arXiv:2312.11805 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  2. [2]

    Qwen3 Technical Report

    Qwen3 technical report , author=. arXiv preprint arXiv:2505.09388 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  3. [3]

    Proceedings of the AAAI Conference on Artificial Intelligence , year=

    Audiogpt: Understanding and generating speech, music, sound, and talking head , author=. Proceedings of the AAAI Conference on Artificial Intelligence , year=

    work page

  4. [4]

    Qwen-Audio: Advancing Universal Audio Understanding via Unified Large-Scale Audio-Language Models

    Qwen-audio: Advancing universal audio understanding via unified large-scale audio-language models , author=. arXiv preprint arXiv:2311.07919 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  5. [5]

    Santa Clara Univ

    An introduction to the california consumer privacy act (ccpa) , author=. Santa Clara Univ. Legal Studies Research Paper , year=

    work page

  6. [6]

    A Practical Guide, 1st Ed., Cham: Springer International Publishing , year=

    The eu general data protection regulation (gdpr) , author=. A Practical Guide, 1st Ed., Cham: Springer International Publishing , year=

    work page

  7. [7]

    for now , author=

    To generate or not? safety-driven unlearned diffusion models are still easy to generate unsafe images... for now , author=. European Conference on Computer Vision , year=

    work page

  8. [8]

    Unlearning or Concealment? A Critical Analysis and Evaluation Metrics for Unlearning in Diffusion Models

    Unlearning or concealment? a critical analysis and evaluation metrics for unlearning in diffusion models , author=. arXiv preprint arXiv:2409.05668 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  9. [9]

    UnlearnCanvas: Stylized Image Dataset for Enhanced Machine Unlearning Evaluation in Diffusion Models

    Unlearncanvas: A stylized image dataset to benchmark machine unlearning for diffusion models , author=. arXiv preprint arXiv:2402.11846 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  10. [10]

    Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Models

    Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Models , author=. arXiv preprint arXiv:2405.15234 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  11. [11]

    Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems , year=

    Are deepfakes concerning? analyzing conversations of deepfakes on reddit and exploring societal implications , author=. Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems , year=

    work page 2022

  12. [12]

    And he’s not happy about it , author=

    This artist is dominating AI-generated art. And he’s not happy about it , author=. MIT Technology Review , year=

    work page

  13. [13]

    32nd USENIX Security Symposium (USENIX Security 23) , year=

    Extracting training data from diffusion models , author=. 32nd USENIX Security Symposium (USENIX Security 23) , year=

    work page

  14. [14]

    32nd USENIX Security Symposium (USENIX Security 23) , year=

    Glaze: Protecting artists from style mimicry by \ Text-to-Image \ models , author=. 32nd USENIX Security Symposium (USENIX Security 23) , year=

    work page

  15. [15]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , year=

    Diffusion art or digital forgery? investigating data replication in diffusion models , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , year=

    work page

  16. [16]

    Foundations and trends

    Advances and open problems in federated learning , author=. Foundations and trends

    work page

  17. [17]

    ACM computing surveys (CSUR) , year=

    The creation and detection of deepfakes: A survey , author=. ACM computing surveys (CSUR) , year=

    work page

  18. [18]

    Foreign Aff

    Deepfakes and the new disinformation war: The coming age of post-truth geopolitics , author=. Foreign Aff. , year=

    work page

  19. [19]

    30th USENIX Security Symposium (USENIX Security 21) , year=

    Extracting training data from large language models , author=. 30th USENIX Security Symposium (USENIX Security 21) , year=

    work page

  20. [20]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , year=

    Vector quantized diffusion model for text-to-image synthesis , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , year=

    work page

  21. [21]

    Advances in Neural Information Processing Systems , year=

    Structured denoising diffusion models in discrete state-spaces , author=. Advances in Neural Information Processing Systems , year=

    work page

  22. [22]

    Hierarchical Text-Conditional Image Generation with CLIP Latents

    Hierarchical text-conditional image generation with clip latents , author=. arXiv preprint arXiv:2204.06125 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  23. [23]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , year=

    Taming transformers for high-resolution image synthesis , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , year=

    work page

  24. [24]

    Advances in neural information processing systems , year=

    Score-based generative modeling in latent space , author=. Advances in neural information processing systems , year=

    work page

  25. [25]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , year=

    High-resolution image synthesis with latent diffusion models , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , year=

    work page

  26. [26]

    Advances in neural information processing systems , year=

    Maximum likelihood training of score-based diffusion models , author=. Advances in neural information processing systems , year=

    work page

  27. [27]

    Advances in neural information processing systems , year=

    Generative modeling by estimating gradients of the data distribution , author=. Advances in neural information processing systems , year=

    work page

  28. [28]

    Score-Based Generative Modeling through Stochastic Differential Equations

    Score-based generative modeling through stochastic differential equations , author=. arXiv preprint arXiv:2011.13456 , year=

    work page internal anchor Pith review Pith/arXiv arXiv 2011

  29. [29]

    Advances in neural information processing systems , year=

    Denoising diffusion probabilistic models , author=. Advances in neural information processing systems , year=

    work page

  30. [30]

    International conference on machine learning , year=

    Deep unsupervised learning using nonequilibrium thermodynamics , author=. International conference on machine learning , year=

    work page

  31. [31]

    Locality Alignment Improves Vision-Language Models

    Locality Alignment Improves Vision-Language Models , author=. arXiv preprint arXiv:2410.11087 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  32. [32]

    LaMP: Language-Motion Pretraining for Motion Generation, Retrieval, and Captioning

    LaMP: Language-Motion Pretraining for Motion Generation, Retrieval, and Captioning , author=. arXiv preprint arXiv:2410.07093 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  33. [33]

    A Dataset and Benchmark for Copyright Infringement Unlearning from Text-to-Image Diffusion Models

    A Dataset and Benchmark for Copyright Protection from Text-to-Image Diffusion Models , author=. arXiv preprint arXiv:2403.12052 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  34. [34]

    Can We Talk Models Into Seeing the World Differently?

    Are Vision Language Models Texture or Shape Biased and Can We Steer Them? , author=. arXiv preprint arXiv:2403.09193 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  35. [35]

    Privacy-Aware Visual Language Models

    Privacy-Aware Visual Language Models , author=. arXiv preprint arXiv:2405.17423 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  36. [36]

    End-to-end Knowledge Retrieval with Multi-modal Queries

    End-to-end knowledge retrieval with multi-modal queries , author=. arXiv preprint arXiv:2306.00424 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  37. [37]

    European conference on computer vision , year=

    Vqa-lol: Visual question answering under the lens of logic , author=. European conference on computer vision , year=

    work page

  38. [38]

    Advances in Neural Information Processing Systems , year=

    Multimodal contrastive learning with limoe: the language-image mixture of experts , author=. Advances in Neural Information Processing Systems , year=

    work page

  39. [39]

    International conference on machine learning , year=

    Vilt: Vision-and-language transformer without convolution or region supervision , author=. International conference on machine learning , year=

    work page

  40. [40]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , year=

    Clippo: Image-and-language understanding from pixels only , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , year=

    work page

  41. [41]

    CoCa: Contrastive Captioners are Image-Text Foundation Models

    Coca: Contrastive captioners are image-text foundation models , author=. arXiv preprint arXiv:2205.01917 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  42. [42]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , year=

    Flava: A foundational language and vision alignment model , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , year=

    work page

  43. [43]

    Supervision Exists Everywhere: A Data Efficient Contrastive Language-Image Pre-training Paradigm

    Supervision exists everywhere: A data efficient contrastive language-image pre-training paradigm , author=. arXiv preprint arXiv:2110.05208 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  44. [44]

    FILIP: Fine-grained Interactive Language-Image Pre-Training

    Filip: Fine-grained interactive language-image pre-training , author=. arXiv preprint arXiv:2111.07783 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  45. [45]

    Advances in Neural Information Processing Systems , year=

    Uniclip: Unified framework for contrastive language-image pre-training , author=. Advances in Neural Information Processing Systems , year=

    work page

  46. [46]

    Proceedings of the AAAI Conference on Artificial Intelligence , year=

    Unifying vision-language representation space with single-tower transformer , author=. Proceedings of the AAAI Conference on Artificial Intelligence , year=

    work page

  47. [47]

    IEEE Transactions on Pattern Analysis and Machine Intelligence , year=

    Vision-language models for vision tasks: A survey , author=. IEEE Transactions on Pattern Analysis and Machine Intelligence , year=

    work page

  48. [48]

    Machine Intelligence Research , year=

    Vlp: A survey on vision-language pre-training , author=. Machine Intelligence Research , year=

    work page

  49. [49]

    International conference on machine learning , year=

    Learning transferable visual models from natural language supervision , author=. International conference on machine learning , year=

    work page

  50. [50]

    International conference on machine learning , year=

    Scaling up visual and vision-language representation learning with noisy text supervision , author=. International conference on machine learning , year=

    work page

  51. [51]

    Advances in neural information processing systems , year=

    Align before fuse: Vision and language representation learning with momentum distillation , author=. Advances in neural information processing systems , year=

    work page

  52. [52]

    European Union Agency for Fundamental Rights , title =

    work page

  53. [53]

    Using Interleaved Ensemble Unlearning to Keep Backdoors at Bay for Finetuning Vision Transformers

    Using Interleaved Ensemble Unlearning to Keep Backdoors at Bay for Finetuning Vision Transformers , author=. arXiv preprint arXiv:2410.01128 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  54. [54]

    Can Small Language Models Learn, Unlearn, and Retain Noise Patterns?

    Can Small Language Models Learn, Unlearn, and Retain Noise Patterns? , author=. arXiv preprint arXiv:2407.00996 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  55. [55]

    AND: Audio Network Dissection for Interpreting Deep Acoustic Models

    AND: Audio Network Dissection for Interpreting Deep Acoustic , author=. arXiv preprint arXiv:2406.16990 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  56. [56]

    Large Scale Knowledge Washing

    Large Scale Knowledge Washing , author=. arXiv preprint arXiv:2405.16720 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  57. [57]

    Guardrail Baselines for Unlearning in LLMs

    Guardrail baselines for unlearning in llms , author=. arXiv preprint arXiv:2403.03329 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  58. [58]

    Erasing Conceptual Knowledge from Language Models

    Erasing Conceptual Knowledge from Language Models , author=. arXiv preprint arXiv:2410.02760 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  59. [59]

    Generative Adversarial Networks Unlearning

    Generative adversarial networks unlearning , author=. arXiv preprint arXiv:2308.09881 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  60. [60]

    Proceedings of the AAAI Conference on Artificial Intelligence (AAAI) , year =

    Saemi Moon and Seunghyuk Cho and Dongwoo Kim , title =. Proceedings of the AAAI Conference on Artificial Intelligence (AAAI) , year =

    work page

  61. [61]

    Receler: Reliable Concept Erasing of Text-to-Image Diffusion Models via Lightweight Erasers

    Receler: Reliable concept erasing of text-to-image diffusion models via lightweight erasers , author=. arXiv preprint arXiv:2311.17717 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  62. [62]

    The Twelfth International Conference on Learning Representations , year=

    Machine unlearning for image-to-image generative models , author=. The Twelfth International Conference on Learning Representations , year=

    work page

  63. [63]

    ConceptPrune: Concept Editing in Diffusion Models via Skilled Neuron Pruning

    ConceptPrune: Concept Editing in Diffusion Models via Skilled Neuron Pruning , author=. arXiv preprint arXiv:2405.19237 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  64. [64]

    Proceedings of International Conference on Computer Vision (ICCV) , month =

    Deep Learning Face Attributes in the Wild , author =. Proceedings of International Conference on Computer Vision (ICCV) , month =

    work page

  65. [65]

    A Survey of Machine Unlearning

    A survey of machine unlearning , author=. arXiv preprint arXiv:2209.02299 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  66. [66]

    Information & Communications Technology Law , year=

    The European Union general data protection regulation: what it is and what it means , author=. Information & Communications Technology Law , year=

    work page

  67. [67]

    Knowledge Unlearning for LLMs: Tasks, Methods, and Challenges

    Knowledge unlearning for llms: Tasks, methods, and challenges , author=. arXiv preprint arXiv:2311.15766 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  68. [68]

    Artificial Intelligence Review , year=

    Digital forgetting in large language models: A survey of unlearning methods , author=. Artificial Intelligence Review , year=

    work page

  69. [69]

    Machine Unlearning in Generative AI: A Survey

    Machine unlearning in generative ai: A survey , author=. arXiv preprint arXiv:2407.20516 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  70. [70]

    Nature Machine Intelligence , year=

    Rethinking machine unlearning for large language models , author=. Nature Machine Intelligence , year=

    work page

  71. [71]

    A Survey on Generative Model Unlearning: Fundamentals, Taxonomy, Evaluation, and Future Direction

    A survey on generative model unlearning: Fundamentals, taxonomy, evaluation, and future direction , author=. arXiv preprint arXiv:2507.19894 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  72. [72]

    A Comprehensive Survey of Machine Unlearning Techniques for Large Language Models

    A comprehensive survey of machine unlearning techniques for large language models , author=. arXiv preprint arXiv:2503.01854 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  73. [73]

    2015 IEEE symposium on security and privacy , year=

    Towards making systems forget with machine unlearning , author=. 2015 IEEE symposium on security and privacy , year=

    work page 2015

  74. [74]

    Advances in neural information processing systems , year=

    Making ai forget you: Data deletion in machine learning , author=. Advances in neural information processing systems , year=

    work page

  75. [75]

    International Conference on Machine Learning , year=

    Certified Data Removal from Machine Learning Models , author=. International Conference on Machine Learning , year=

    work page

  76. [76]

    2021 IEEE Symposium on Security and Privacy (SP) , year=

    Machine unlearning , author=. 2021 IEEE Symposium on Security and Privacy (SP) , year=

    work page 2021

  77. [77]

    Advances in Neural Information Processing Systems , year=

    Remember what you want to forget: Algorithms for machine unlearning , author=. Advances in Neural Information Processing Systems , year=

    work page

  78. [78]

    Algorithmic Learning Theory , year=

    Descent-to-delete: Gradient-based methods for machine unlearning , author=. Algorithmic Learning Theory , year=

    work page

  79. [79]

    Theory of Cryptography: Third Theory of Cryptography Conference, TCC 2006, New York, NY, USA, March 4-7, 2006

    Calibrating noise to sensitivity in private data analysis , author=. Theory of Cryptography: Third Theory of Cryptography Conference, TCC 2006, New York, NY, USA, March 4-7, 2006. Proceedings 3 , year=

    work page 2006

  80. [80]

    Are we making progress in unlearning? Findings from the first NeurIPS unlearning competition

    Are we making progress in unlearning? Findings from the first NeurIPS unlearning competition , author=. arXiv preprint arXiv:2406.09073 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

Showing first 80 references.