您當(dāng)前的位置：檢測資訊 > 法規(guī)標準
嘉峪檢測網(wǎng) 2025-07-09 21:41
7月7日，歐盟委員會和PIC/S均發(fā)布了新的GMP修訂：包括正文第4章《文件記錄》、附錄11《計算機化系統(tǒng)》和一個新的附錄——附錄22《人工智能》：
附錄22為首個人工智能相關(guān)GMP指南，文件反復(fù)強調(diào)了不應(yīng)在關(guān)鍵GMP應(yīng)用中使用動態(tài)模型，生成式人工智能和大型語言模型（LLM）和具有概率性輸出的模型。以下為附錄22《人工智能》中英文對照：
Annex 22: Artificial Intelligence附錄 22：人工智能
Reasons for changes: Not applicable (new annex).變更原因：新增附錄
Document map文件目錄
1. Scope范圍
2. Principles原則
3. Intended Use預(yù)期用途
4. Acceptance Criteria接受標準5. Test Data測試數(shù)據(jù)
6. Test Data Independency測試數(shù)據(jù)獨立性
7. Test Execution測試執(zhí)行
8. Explainability可解釋性
9. Confidence置信度
10. Operation運行（操作） Glossary術(shù)語表
1.Scope范圍
This annex applies to all types of computerised systems used in the manufacturing of medicinal products and active substances, where Artificial Intelligence models are used in critical applications with direct impact on patient safety, product quality or data integrity, e.g. to predict or classify data. The document provides additional guidance to Annex11 for computerised systems in which AI models are embedded.本附錄適用于藥品和活性物質(zhì)生產(chǎn)中使用人工智能模型用于對患者安全、產(chǎn)品質(zhì)量或數(shù)據(jù)完整性有直接影響的關(guān)鍵應(yīng)用（例如用于數(shù)據(jù)預(yù)測或分類）的各類計算機化系統(tǒng)。對于嵌入了人工智能模型的計算機化系統(tǒng)，本文件為《附錄11》提供了補充指導(dǎo)。
The document applies to machine learning (AI/ML) models which have obtained their functionality through training with data, rather than being explicitly programmed. Models may consist of several individual models, each automating specific process steps in GMP.本文件適用于通過數(shù)據(jù)訓(xùn)練（而非通過明確編程）而非通過明確編程獲得功能的機器學(xué)習(xí)（人工智能/機器學(xué)習(xí)）模型。這些模型可由多個獨立模型構(gòu)成，每個模型負責(zé)自動化處理GMP中的特定流程步驟。
The document applies to static models, i.e. models that do not adapt their performance during use by incorporating new data. The use of dynamic models which continuously and automatically learn and adapt performance during use, is not covered by this document, and should not be used in critical GMP applications.本文件適用于靜態(tài)模型，即那些在使用過程中不會通過納入新數(shù)據(jù)來調(diào)整自身性能的模型。而動態(tài)模型在使用過程中會持續(xù)自動學(xué)習(xí)并調(diào)整性能，其使用不在本文件的涵蓋范圍內(nèi)，且不應(yīng)在關(guān)鍵GMP應(yīng)用中使用。
The document applies to models with a deterministic output which, when given identical inputs, provide identical outputs. Models with a probabilistic output which, when given identical inputs, might not provide identical outputs are not covered by this document and should not be used in critical GMP applications.本文件適用于具有確定性輸出的模型，即當(dāng)輸入相同時，這類模型能給出相同的輸出。而具有概率性輸出的模型在輸入相同時可能不會產(chǎn)生相同的輸出，此類模型不在本文件的涵蓋范圍內(nèi)，且不應(yīng)在關(guān)鍵GMP應(yīng)用中使用。
Following the above, the document does not apply to Generative AI and Large Language Models (LLM), and such models should not be used in critical GMP applications. If used in non-critical GMP applications, which do not have direct impact on patient safety, product quality or data integrity, personnel with adequate qualification and training should always be responsible for ensuring that the outputs from such models are suitable for the intended use, i.e. a human-in-the-loop (HITL) and the principles described in this document may be considered where applicable.基于上述內(nèi)容，本文件不適用于生成式人工智能和大型語言模型（LLM），且此類模型不應(yīng)在關(guān)鍵GMP應(yīng)用中使用。若在非關(guān)鍵的GMP應(yīng)用（即對患者安全、產(chǎn)品質(zhì)量或數(shù)據(jù)完整性無直接影響的應(yīng)用）中使用這類模型，應(yīng)由具備適當(dāng)資質(zhì)和培訓(xùn)的人員負責(zé)確保模型輸出符合預(yù)期用途，即需采用“人機協(xié)同（HITL）”模式，且在適用情況下可參考本文件所述原則。
2. Principles2. 原則
2.1.Personnel. 2.1. 人員
In order to adequately understand the intended use and the associated risks of the application of an AI model in a GMP environment, there should be close cooperation between all relevant parties during algorithm selection, and model training, validation, testing and operation. This includes but may not be limited to process subject matter experts (SMEs), QA, data scientists, IT, and consultants. All personnel should have adequate qualifications, defined responsibilities and appropriate level of access.為充分理解人工智能模型在GMP環(huán)境中應(yīng)用的預(yù)期用途及相關(guān)風(fēng)險，在算法選擇、模型訓(xùn)練、驗證、測試和運行期間，所有相關(guān)方應(yīng)開展密切合作。相關(guān)方包括但不限于工藝主題專家（SMEs）、QA、數(shù)據(jù)科學(xué)家、IT、和顧問。所有人員均應(yīng)具備適當(dāng)資質(zhì)、明確的職責(zé)和相應(yīng)的訪問權(quán)限。
2.2.Documentation. 2.2. 文件記錄
Documentation for activities described in this section should be available and reviewed by the regulated user irrespective of whether a model is trained, validated and tested in-house or whether it is provided by a supplier or service provider.無論模型是內(nèi)部訓(xùn)練、驗證和測試的，還是由供應(yīng)商或服務(wù)提供商提供，應(yīng)有本節(jié)所述活動的文件記錄，并經(jīng)受監(jiān)管用戶審核。
2.3.Quality Risk Management質(zhì)量風(fēng)險管理
Activities described in this document should be implemented based on the risk to patient safety, product quality and data integrity.本文件所述活動應(yīng)基于患者安全、產(chǎn)品質(zhì)量和數(shù)據(jù)完整性的風(fēng)險實施。
3. Intended Use3. 預(yù)期用途
3.1.Intended use. The intended use of a model and the specific tasks it is designed to assist or automate should be described in detail based on an in-depth knowledge of the process the model is integrated in. This should include a comprehensive characterisation of the data the model is intended to use as input and all common and rare variations; i.e. the input sample space. Any limitations and possible erroneous and biased inputs should be identified. A process subject matter expert (SME) should be responsible for the adequacy of the description, and it should be documented and approved before the start of acceptance testing.3.1. 預(yù)期用途：應(yīng)基于對模型所融入的流程的深入了解，詳細描述模型的預(yù)期用途以及其設(shè)計用于輔助或自動化的特定任務(wù)。這應(yīng)包括對模型擬使用的輸入數(shù)據(jù)、所有常見及罕見的數(shù)據(jù)變異（即輸入樣本空間）的全面特征分析。此外，還應(yīng)識別出模型的任何局限性以及可能存在的錯誤輸入和有偏輸入。流程領(lǐng)域?qū)＜遥⊿ME）需對上述描述的充分性負責(zé)，且該描述需在驗收測試開始前形成文件并獲得批準。
3.2.Subgroups. Where applicable, the input sample space should be divided into subgroups based on relevant characteristics. Subgroups may be defined by characteristics like the decision output (e.g. ‘accept’ or ‘reject’), process specific baseline characteristics (e.g. geographical site or equipment), specific characteristics in material or product, and characteristics specific to the task being automated (e.g. types and severity of defects).3.2. 子組：在適用情況下，應(yīng)根據(jù)相關(guān)特征將輸入樣本空間劃分為子組。子組可通過以下特征來定義：決策輸出（如“接受”或“拒絕”）、特定工藝基準特征（如生產(chǎn)地點或設(shè)備）、物料或產(chǎn)品的特定屬性，以及自動化任務(wù)所特有的特征（如缺陷類型和嚴重程度）。
3.3.Human-in-the-loop. Where a model is used to give an input to a decision made by a human operator (human-in-the-loop), and where the effort to test such model has been diminished, the description ofthe intended use should include theresponsibility of the operator. In this case, the training and consistent performance of the operator should be monitored like any other manual process.3.3. 人機協(xié)同：當(dāng)模型用于為人工操作員的決策提供輸入（即人機協(xié)同），且對該模型的測試力度有所降低時，其預(yù)期用途描述應(yīng)包含操作員的職責(zé)。在此情況下，操作員的培訓(xùn)情況及持續(xù)表現(xiàn)應(yīng)像其他任何人工流程一樣受到監(jiān)控。
4. Acceptance Criteria4. 接受標準
4.1.Test metrics. Suitable, case dependent test metrics, should be defined to measure the performance of the model according to the intended use. As an example, suitable test metrics for a model used to classify products (e.g. ‘accept’ or ‘reject’ ) may include, but may not be limited to, a confusion matrix, sensitivity, specificity, accuracy, precision and/or F1 score.4.1 測試指標：應(yīng)根據(jù)具體情況，定義適當(dāng)?shù)臏y試指標，以依據(jù)預(yù)期用途衡量模型的性能。例如，對于用于產(chǎn)品分級（如“接受”或“拒絕”）的模型，合適的測試指標可包括但不限于混淆矩陣、靈敏度、特異度、準確率、精確率和/或F1分數(shù)。
4.2.Acceptance criteria. Acceptance criteria for the defined test metrics should be established by which the performance of the model should be considered acceptable for the intended use. The acceptance criteria may differ for specific subgroups within the intended use. A process subject matter expert (SME) should be responsible for the definition of the acceptance criteria, which should be documented and approved before the start of acceptance testing.4.2. 接受標準：應(yīng)針對所定義的測試指標制定接受標準，以此判定模型性能是否滿足預(yù)期用途要求。預(yù)期用途中的特定子組可能適用不同的接受標準。流程主題專家（SME）應(yīng)負責(zé)定義接受標準，該標準應(yīng)形成文件并在驗收測試開始前獲得批準。
4.3.No decrease. The acceptance criteria of a model, should be at least as high as the performance of the process it replaces. This implies, that the performance should be known for the process which is to be replaced by a model (see Annex 11 2.7).4.3. 不降低要求：模型的接受標準應(yīng)至少不低于其所要替代的流程的性能水平。這意味著，對于擬由模型替代的流程，其性能水平應(yīng)是已知的（參見《附錄 11》第 2.7 條）。
5. Test Data5. 測試數(shù)據(jù)
5.1.Selection. Test data should be representative of and expand the full sample space of the intended use. It should be stratified, include all subgroups, and reflect the limitations, complexity and all common and rare variations within the intended use of the model. The criteria and rationale for selection of test data should be documented.5.1. 選擇：測試數(shù)據(jù)應(yīng)能代表并覆蓋預(yù)期用途的完整樣本空間。測試數(shù)據(jù)應(yīng)經(jīng)過分層處理，涵蓋所有子組，并能反映模型預(yù)期用途范圍內(nèi)的局限性、復(fù)雜性以及所有常見和罕見的變異情況。應(yīng)記錄測試數(shù)據(jù)選擇的標準和理由。
5.2.Sufficient in size. The test dataset, and any of its subgroups, should be sufficient in size to calculate the test metrics with adequate statistical confidence.5.2. 規(guī)模充足：測試數(shù)據(jù)集及其包含的所有子組，在規(guī)模上應(yīng)足以使測試指標的計算具備充分的統(tǒng)計置信度。
5.3.Labelling. The labelling of test data should be verified following a process that ensures a very high degree of correctness. This may include independent verification by multiple experts, validated equipment or laboratory tests.5.3. 標記：測試數(shù)據(jù)的標記應(yīng)通過能確保極高正確性的流程進行驗證。這可包括由多名專家進行獨立驗證、使用經(jīng)過驗證的設(shè)備驗證或通過實驗室檢測驗證。
5.4.Pre-processing. Any pre-processing of the test data, e.g. transformation, normalisation, or standardisation, should be pre-specified and a rationale should be provided, that it represents intended use conditions.5.4. 預(yù)處理：對測試數(shù)據(jù)的任何預(yù)處理（如轉(zhuǎn)換、歸一化或標準化）都應(yīng)預(yù)先規(guī)定，并提供相應(yīng)理由，以證明其符合預(yù)期使用條件。
5.5.Exclusion. Any cleaning or exclusion of test data should be documented and fully justified.5.5. 排除：對測試數(shù)據(jù)的任何清洗或排除操作均應(yīng)形成文件記錄，并提供充分的理由。
5.6.Data generation. Generation of test data or labels, e.g. by means of generative AI, is not recommended and any use hereof should be fully justified.5.6. 數(shù)據(jù)生成：不建議通過生成式人工智能等方式生成測試數(shù)據(jù)或標記，若確需使用此類生成的數(shù)據(jù)或標記，必須提供充分的理由。
6. Test Data Independency6. 測試數(shù)據(jù)獨立性
6.1.Independence. Effective measures consisting of technical and/or procedural controls should be implemented to ensure the independency of test data, i.e. that data which will be used to test a model, is not used during development, training or validation of the model. This may be by capturing test data only after completion of training and validation, or by splitting test data from a complete pool of data before training has started.6.1. 獨立性：應(yīng)實施由技術(shù)和/或程序控制組成的有效措施，確保測試數(shù)據(jù)的獨立性，即確保用于測試模型的數(shù)據(jù)未在模型的開發(fā)、訓(xùn)練或驗證階段使用。這可以通過僅在訓(xùn)練和驗證完成后收集測試數(shù)據(jù)，或在訓(xùn)練開始前從完整數(shù)據(jù)池中分離出測試數(shù)據(jù)來實現(xiàn)。
6.2.Data split. If test data is split from a complete pool of data before training of the model, it is essential that employees involved in the development and training of the model have never had access to the test data. The test data should be protected by access control and audit trail functionality logging accesses and changes to these. There should be no copies of test data outside this repository.6.2. 數(shù)據(jù)拆分：若在模型訓(xùn)練前從完整數(shù)據(jù)池中拆分出測試數(shù)據(jù)，至關(guān)重要的一點是：參與模型開發(fā)和訓(xùn)練的人員絕不能接觸到測試數(shù)據(jù)。測試數(shù)據(jù)應(yīng)通過訪問控制和審計追蹤功能加以保護，以記錄對測試數(shù)據(jù)的訪問和修改情況。此外，該數(shù)據(jù)存儲庫之外不應(yīng)存在測試數(shù)據(jù)的副本。
6.3.Identification. It should be recorded which data has been used for testing, when and how many times.6.3. 標識：應(yīng)記錄哪些數(shù)據(jù)用于測試、測試的時間以及測試的次數(shù)。
6.4.Physical objects. When test data originates from physical objects, it should be ensured, that the objects used for the final test of the model have not previously been used to train or validate the model, unless features are independent.6.4. 實物對象：當(dāng)測試數(shù)據(jù)源自實物對象時，應(yīng)確保用于模型最終測試的實物對象此前未被用于模型的訓(xùn)練或驗證，除非其特征具有獨立性。
6.5.Staff independency. Effective procedural and/or technical controls should be implemented to prevent staff members who have had access to test data from being involved in training and validation of the same model. In organisations where it is impossible to maintain this independency, a staff member who might have had access to test data for a model, should only have access to training and validation of the same model when working together (in pair) with a colleague who has not had this access (4-eyes principle).6.5. 人員獨立性：應(yīng)實施有效的程序和/或技術(shù)控制措施，防止接觸過測試數(shù)據(jù)的人員參與同一模型的訓(xùn)練和驗證工作。在無法保持這種獨立性的組織中，可能接觸過某模型測試數(shù)據(jù)的人員，只有在與未接觸過該測試數(shù)據(jù)的同事共同工作（雙人協(xié)作）時，方可參與同一模型的訓(xùn)練和驗證（即遵循“四眼原則”）。
7. Test Execution7. 測試執(zhí)行
7.1.Fit for intended use. The test should ensure that a model is fit for intended use and is ‘generalising well’, i.e. that the model has a satisfactory performance with new data from the intended use. This includes detecting possible over- or underfitting of the model to the training data.7.1. 符合預(yù)期用途：測試應(yīng)確保模型符合預(yù)期用途且“泛化能力良好”，即模型在處理來自預(yù)期用途的新數(shù)據(jù)時表現(xiàn)令人滿意。這包括檢測模型對訓(xùn)練數(shù)據(jù)可能存在的過擬合或欠擬合問題。
7.2.Testplan. Before the test is initiated, a test plan should be prepared and approved. It should contain a summary of the intended use, the pre-defined metrics and acceptance criteria, a reference to the test data, a test script including a description of all steps necessary to conduct the test, and a description of how to calculate the test metrics. A process subject matter expert (SME) should be involved in developing the plan.7.2. 測試計劃：在啟動測試前，應(yīng)編制并批準測試計劃。該計劃應(yīng)包含以下內(nèi)容：預(yù)期用途概述、預(yù)先定義的指標和接受標準、測試數(shù)據(jù)的引用、包含實施測試所需全部步驟說明的測試腳本，以及測試指標的計算方法說明。流程主題專家（SME）應(yīng)參與測試計劃的制定。
7.3.Deviation. Any deviation from the test plan, failure to meet acceptance criteria, or omission to use all test data should be documented, investigated, and fully justified.7.3. 偏差：任何與測試計劃不符的情況、未達到接受標準的情況，或未使用全部測試數(shù)據(jù)的情況，都應(yīng)記錄、調(diào)查并充分論證。
7.4.Test documentation. All test documentation should be retained along with the description of the intended use, the characterisation of test data, the actual test data, and whererelevant, physical test objects. In addition, documentation for access control to test data and related audit trail records, should be retained similarly to other GMP documentation.7.4. 測試文件記錄：所有測試文件記錄均應(yīng)留存，同時留存的還應(yīng)包括預(yù)期用途說明、測試數(shù)據(jù)特征描述、實際測試數(shù)據(jù)，以及相關(guān)情況下的實物測試對象。此外，測試數(shù)據(jù)的訪問控制文件記錄及相關(guān)審計追蹤記錄，應(yīng)與其他GMP文件記錄一樣妥善留存。
8. Explainability8. 可解釋性
8.1.Feature attribution. During testing of models used in critical GMP applications, systems should capture and record the features in the test data that have contributed to a particular classification or decision (e.g. rejection). Where applicable, techniques like feature attribution (e.g. SHAP values or LIME) or visual tools like heat maps should be used to highlight key factors contributing to the outcome.8.1. 特征歸因：在對關(guān)鍵GMP應(yīng)用中使用的模型進行測試時，系統(tǒng)應(yīng)捕捉并記錄測試數(shù)據(jù)中促成特定分類或決策（如拒收）的特征。在適用的情況下，應(yīng)使用特征歸因技術(shù)（如SHAP值或LIME）或熱圖等可視化工具，突出顯示導(dǎo)致結(jié)果的關(guān)鍵因素。
8.2.Feature justification. In order to ensure that a model is making decisions based on relevant and appropriate features and based on risk, a review of these features should be part of the process for approval of test results.8.2. 特征合理性論證：為確保模型基于相關(guān)且適當(dāng)?shù)奶卣饕约帮L(fēng)險來制定決策，對這些特征的審核應(yīng)成為測試結(jié)果審批流程的一部分。
9. Confidence9. 置信度
9.1.Confidencescore. When testing a model used to predict or classify data,the system should, where applicable, log the confidence score of the model for each prediction or classification outcome.9.1. 置信度分數(shù)：在測試用于預(yù)測或分類數(shù)據(jù)的模型時，系統(tǒng)應(yīng)在適用情況下記錄模型對每個預(yù)測或分類結(jié)果的置信度分數(shù)。
9.2.Threshold. Models used to predict or classify data should have an appropriate threshold setting to ensure predictions or classifications are made only when suitable. If the confidence score is very low, it should be considered whether the model should flag the outcome as ‘undecided’, rather than making potentially unreliable predictions or classifications.9.2. 閾值：用于預(yù)測或分類數(shù)據(jù)的模型應(yīng)設(shè)置適當(dāng)?shù)拈撝?，以確保僅在合適的情況下進行預(yù)測或分類。若置信度分數(shù)極低，則應(yīng)考慮模型是否應(yīng)將結(jié)果標記為“未確定”，而非做出可能不可靠的預(yù)測或分類。
10. Operation10. 運行
10.1.Change control. A tested model, the system it is implemented in, and the whole process it is automating or assisting should be put under change control before it is deployed in operation. Any change to the model itself, the system, or the process in which it is used, including any change to physical objects the model is using as input, should be documented and evaluated to determine if the model needs to be retested. Any decision not to conduct such retest should be fully justified.10.1. 變更控制：經(jīng)測試的模型、其部署的系統(tǒng)以及它所自動化或輔助的整個流程，在投入運行前均應(yīng)納入變更控制范圍。對模型本身、系統(tǒng)或其應(yīng)用流程的任何變更（包括模型用作輸入的實物對象的任何變更），都應(yīng)形成文件記錄并進行評估，以確定是否需要對模型重新測試。對于決定不進行此類重新測試的情況，必須提供充分的理由。
10.2.Configuration control. A tested model should be put under configuration control before being deployed in operation, and effective measures should be used to detect any unauthorised change.10.2. 配置控制：經(jīng)過測試的模型在投入運行前應(yīng)納入配置控制范圍，并應(yīng)采取有效措施檢測任何未經(jīng)授權(quán)的變更。
10.3.System performance monitoring. The performance of a model as defined by its metrics should be regularly monitored to detect any changes in the computerised system (e.g. deterioration or change of a lighting condition).10.3. 系統(tǒng)性能監(jiān)控：應(yīng)定期監(jiān)控模型在其指標所定義的性能表現(xiàn)，以檢測計算機化系統(tǒng)中出現(xiàn)的任何變化（例如光照條件的惡化或改變）。
10.4.Input sample space monitoring. It should be regularly monitored whether the input data are still within the model sample space and intended use. Metrics should be defined for monitoring any drift in the input data.
10.4. 輸入樣本空間監(jiān)控：應(yīng)定期監(jiān)控輸入數(shù)據(jù)是否仍處于模型樣本空間及預(yù)期用途范圍內(nèi)。應(yīng)為監(jiān)控輸入數(shù)據(jù)的任何偏移情況定義相關(guān)指標。
10.5.Human review. When a model is used to give an input to a decision made by a human operator (human-in-the-loop), and where the effort to test such model has been diminished, records should be kept from this process. Depending on the criticality of the process and the level of testing of the model, this may imply a consistent review and/or test of every output from the model, according to a procedure.
10.5. 人工審核：當(dāng)模型用于為人工操作員的決策提供輸入（即“人機協(xié)同”模式），且對此類模型的測試力度有所減弱時，該過程需留存記錄。根據(jù)流程的關(guān)鍵程度以及模型的測試水平，這可能意味著需要按照既定程序，對模型的每一項輸出進行持續(xù)審核和/或測試。
Glossary術(shù)語
Artificial Intelligence – ‘AI system’ means a machine-based system that is designed to operate with varying levels of autonomy and that may exhibit adaptiveness after deployment, and that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments;人工智能——“人工智能系統(tǒng)”指一種基于機器的系統(tǒng)，其設(shè)計具有不同程度的自主性，在部署后可能表現(xiàn)出適應(yīng)性；為實現(xiàn)顯性或隱性目標，該系統(tǒng)從接收的輸入中推斷如何生成輸出（如預(yù)測、內(nèi)容、建議或決策），而這些輸出能夠?qū)ξ锢憝h(huán)境或虛擬環(huán)境產(chǎn)生影響。
Deep Learning – Approach to creating rich hierarchical representations through the training of neural networks with many hidden layers深度學(xué)習(xí)——通過訓(xùn)練具有多個隱藏層的神經(jīng)網(wǎng)絡(luò)來創(chuàng)建豐富的層級表示的方法
Feature – A pattern in data that can be reduced to a simpler higher-level representation特征——數(shù)據(jù)中可簡化為更簡潔的高級表示形式的模式
LIME – Local Interpretable Model-Agnostic Explanations; a technique that approximates any black box machine learning model with a local, interpretable model to explain each individual prediction.LIME——局部可解釋的模型無關(guān)解釋（Local Interpretable Model-Agnostic Explanations）；這是一種通過局部可解釋模型來近似任何黑箱機器學(xué)習(xí)模型的技術(shù)，用于解釋每個單獨的預(yù)測結(jié)果。
Machine Learning – Machine learning refers to the computational process of optimising the parameters of a model from data, which is a mathematical construct generating an output based on input data. Machine learning approaches include, for instance, supervised, unsupervised and reinforcement learning, using a variety of methods including deep learning with neural networks.
機器學(xué)習(xí)——機器學(xué)習(xí)是指通過數(shù)據(jù)優(yōu)化模型參數(shù)的計算過程，模型是一種基于輸入數(shù)據(jù)生成輸出的數(shù)學(xué)結(jié)構(gòu)。機器學(xué)習(xí)方法包括（例如）監(jiān)督學(xué)習(xí)、無監(jiān)督學(xué)習(xí)和強化學(xué)習(xí)，采用包括神經(jīng)網(wǎng)絡(luò)深度學(xué)習(xí)在內(nèi)的多種技術(shù)。
Model – Mathematical algorithms with parameters (weights) arranged in an architecture that allows learning of patterns (features) from training data模型——在特定架構(gòu)中排列的帶有參數(shù)（權(quán)重）的數(shù)學(xué)算法，能夠從訓(xùn)練數(shù)據(jù)中學(xué)習(xí)模式（特征）
Overfitting – Learning details from training data that cannot be generalised to new data過擬合——從訓(xùn)練數(shù)據(jù)中學(xué)習(xí)到無法泛化到新數(shù)據(jù)的細節(jié)信息
SHAP – Shapley Additive Explanations; an explainable AI (XAI) framework that can provide model- agnostic local explainability for tabular, image, and text datasetsSHAP——沙普利可加解釋（Shapley Additive Explanations）；這是一種可解釋人工智能（XAI）框架，能夠為表格數(shù)據(jù)、圖像數(shù)據(jù)和文本數(shù)據(jù)集提供與模型無關(guān)的局部可解釋性。
Static – Frozen model: A model where all parameters have been finally set, not allowing further adaption to new data.
靜態(tài)（模型）——凍結(jié)模型：指所有參數(shù)均已最終設(shè)定、不允許進一步根據(jù)新數(shù)據(jù)進行調(diào)整的模型。
Test dataset – The "hold-out" data that is used to estimate performance of the final ML model.
測試數(shù)據(jù)集——用于評估最終機器學(xué)習(xí)模型性能的“預(yù)留”數(shù)據(jù)。
Training dataset – The data used to train the ML model.訓(xùn)練數(shù)據(jù)集——用于訓(xùn)練機器學(xué)習(xí)模型的數(shù)據(jù)。
Validation dataset (in AI) – The dataset used during model development, to inform on how to optimally train the model from training data. size smaller than the training set
驗證數(shù)據(jù)集（在人工智能領(lǐng)域）——模型開發(fā)過程中使用的數(shù)據(jù)集，用于指導(dǎo)如何從訓(xùn)練數(shù)據(jù)中以最優(yōu)方式訓(xùn)練模型。其規(guī)模小于訓(xùn)練集。
來源：Internet
關(guān)于我們 聯(lián)系我們 營業(yè)執(zhí)照 全國服務(wù)電話：400-818-0021 嘉峪檢測版權(quán)所有 備案號：京ICP備14047019號-2 
京公網(wǎng)安備11010802046783號
