Jove
Visualize
お問い合わせ
JoVE
x logofacebook logolinkedin logoyoutube logo
JoVEについて
概要リーダーシップブログJoVEヘルプセンター
著者向け
出版プロセス編集委員会範囲と方針査読よくある質問投稿
図書館員向け
推薦の声購読アクセスリソース図書館諮問委員会よくある質問
研究
JoVE JournalMethods CollectionsJoVE Encyclopedia of Experimentsアーカイブ
教育
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab Manual教員リソースセンター教員サイト
利用規約
プライバシーポリシー
ポリシー

関連する概念動画

Survival Tree01:19

Survival Tree

75
Survival trees are a non-parametric method used in survival analysis to model the relationship between a set of covariates and the time until an event of interest occurs, often referred to as the "time-to-event" or "survival time." This method is particularly useful when dealing with censored data, where the event has not occurred for some individuals by the end of the study period, or when the exact time of the event is unknown.
 Building a Survival Tree
Constructing a...
75
Clearance Models: Compartment Models01:25

Clearance Models: Compartment Models

64
Clearance measures drug elimination from the central compartment, including plasma and highly perfused organs like kidneys and liver. Its calculation varies depending on pharmacokinetic models and administration routes. The one-compartment model, for instance, portrays the pharmacokinetics of polar drugs such as aminoglycoside antibiotics administered intravenously and readily excreted in urine. In this case, clearance is influenced by the terminal rate constant (λz) and the total volume...
64
Mechanistic Models: Compartment Models in Algorithms for Numerical Problem Solving01:29

Mechanistic Models: Compartment Models in Algorithms for Numerical Problem Solving

47
Mechanistic models play a crucial role in algorithms for numerical problem-solving, particularly in nonlinear mixed effects modeling (NMEM). These models aim to minimize specific objective functions by evaluating various parameter estimates, leading to the development of systematic algorithms. In some cases, linearization techniques approximate the model using linear equations.
In individual population analyses, different algorithms are employed, such as Cauchy's method, which uses a...
47
Collisions in Multiple Dimensions: Problem Solving01:06

Collisions in Multiple Dimensions: Problem Solving

3.7K
In multiple dimensions, the conservation of momentum applies in each direction independently. Hence, to solve collisions in multiple dimensions, we should write down the momentum conservation in each direction separately. To help understand collisions in multiple dimensions, consider an example.
A small car of mass 1,200 kg traveling east at 60 km/h collides at an intersection with a truck of mass 3,000 kg traveling due north at 40 km/h. The two vehicles are locked together. What is the...
3.7K
Multicompartment Models: Overview01:14

Multicompartment Models: Overview

121
Multicompartment models are mathematical constructs that depict how drugs are distributed and eliminated within the body. They segment the body into several compartments, symbolizing various physiological or anatomical areas connected through drug transfer processes such as absorption, metabolism, distribution, and elimination.
These models offer a more comprehensive representation of drug behavior in the body than one-compartment models. They accommodate the complexity of drug distribution,...
121
Per-Unit Sequence Models01:26

Per-Unit Sequence Models

73
An ideal Y-Y transformer, grounded through neutral impedances, displays per-unit sequence networks akin to those of a single-phase ideal transformer when subjected to balanced positive- or negative-sequence currents. These currents do not produce neutral currents, and their associated voltage drops.
Zero-sequence currents, which are identical in magnitude and phase, generate a neutral current, resulting in voltage drops across the neutral impedance and the low-voltage winding. If the...
73

こちらも読む

関連記事

共著者、ジャーナル、引用グラフによってこの研究に関連する記事。

並び替え
Same author

Scalable watermarking for identifying large language model outputs.

Nature·2024
Same author

Detecting hallucinations in large language models using semantic entropy.

Nature·2024
Same author

Mining Bodily Cues to Deception.

Journal of nonverbal behavior·2024
Same author

Decentralised, collaborative, and privacy-preserving machine learning for multi-hospital data.

EBioMedicine·2024
Same author

ProteinNPT: Improving Protein Property Prediction and Design with Non-Parametric Transformers.

bioRxiv : the preprint server for biology·2023
Same author

ProteinGym: Large-Scale Benchmarks for Protein Design and Fitness Prediction.

bioRxiv : the preprint server for biology·2023
Same journal

Harmonizing standards and resources for the medical genome.

Nature·2026
Same journal

Towards the construction of a virtual yeast.

Nature·2026
Same journal

Aerosols and hydrocarbons in the atmosphere of a white dwarf planet.

Nature·2026
Same journal

TROP2 targeting reveals therapy-driven cell state dynamics in colorectal cancer.

Nature·2026
Same journal

Competing programs shape cortical sensorimotor-association axis development.

Nature·2026
Same journal

Steatosis shapes prognosis-defining liver metastasis heterogeneity in CRC.

Nature·2026
関連記事をすべて見る

関連する実験動画

Updated: Jun 19, 2025

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
03:14

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

539

AIモデルは再帰的に生成されたデータで訓練されると崩壊します.

Ilia Shumailov1, Zakhar Shumaylov2, Yiren Zhao3

  • 1OATML, Department of Computer Science, University of Oxford, Oxford, UK. ilia.shumailov@chch.ox.ac.uk.

Nature
|July 24, 2024
PubMed
まとめ
この要約は機械生成です。

独創的な人工知能 (AI) のモデルでは,モデル崩壊という現象で,不可逆的な欠陥が発生します. これはAIによって生み出される 未来のコンテンツの質と多様性に 影響を及ぼします

さらに関連する動画

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

2.7K
Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches
09:47

Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches

Published on: December 15, 2023

1.0K

関連する実験動画

Last Updated: Jun 19, 2025

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
03:14

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

539
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

2.7K
Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches
09:47

Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches

Published on: December 15, 2023

1.0K

科学分野:

  • 人工知能
  • 機械学習
  • 生成モデル

背景:

  • GPT-4のような大型言語モデル (LLM) と 安定した拡散のような画像生成モデルを含む 生成型人工知能 (AI) は オンラインコンテンツを急速に変化させています
  • AIで生成されたテキストと画像の普及は,これらのモデルを訓練するために使用されるデータの将来性について疑問を投げかけています.
  • GPT-2,GPT-3,GPT-4などの以前のAIの進歩は,さまざまな言語のタスクにおいて重要な能力を示しています.

研究 の 目的:

  • 大型言語モデル (LLM) が将来のAIトレーニングデータに与える潜在的な影響を調査する.
  • モデルによって生成されたコンテンツでAIモデルが訓練されたときに発生する欠陥を特定し,分析する.
  • これらの欠陥がAI開発の持続可能性と多様なデータソースの価値に及ぼす影響を理解する.

主な方法:

  • LLM,バリエーションオートエンコーダー (VAE),ガウス混合モデル (GMM) を含むジェネラティブモデルの理論分析.
  • モデル崩壊の発生と影響を実証するシミュレーションと経験的研究.
  • 合成データでモデルを訓練すると,データ分布の喪失に関する調査が終了します.

主要な成果:

  • 訓練におけるモデル生成コンテンツの無差別な使用は,AIモデルの不可逆的な欠陥につながります.
  • この現象は"モデル崩壊"と呼ばれ,元のデータ分布の尾の消失を引き起こします.
  • モデル崩壊は,LLM,VAE,GMMを含む様々なタイプの生成モデルにおいて,普遍的な問題であることが示されています.

結論:

  • モデル崩壊はAIによって生み出されるコンテンツの 長期的な品質と多様性に重大な脅威をもたらします
  • 大規模なウェブデータに関するトレーニングの利点を維持するには,モデルの崩壊に対処する必要があります.
  • 人工知能の訓練における モデルの崩壊に対する 対策として リアルな人間のやり取りを反映したデータが 益々価値を持つようになるでしょう