引導式測驗 · 中等 等級
Teachable Machine 教室專案測驗
規劃可教機器項目,教授圖像、聲音和姿勢分類、資料集選擇、測試和安全課堂使用。
問題 1 的 8
哪一個順序符合 Teachable Machine 的基本工作流程?
預覽每個問題8 問題
當您參加測驗時,就會出現答案和解釋。
- 哪一個順序符合 Teachable Machine 的基本工作流程?
- Gather examples, train a model, then test it
- Train on unlabeled data, then decide what classes it should have
- Test examples first, then add labels after prediction
- Export the model before choosing what it should recognize
- 為什麼學生應該測試訓練期間未使用的範例?
- To confirm that training labels are spelled consistently
- To see whether the model handles new inputs
- To make the model memorize examples it has already seen
- To avoid documenting incorrect predictions
- 哪種改變有助於調查模型的失敗原因?
- Change the labels and lighting together without recording either
- Add examples only to a failing class and assume the cause is known
- Vary one condition, such as lighting, while tracking results
- Remove mistaken predictions before analyzing them
- 小型課堂資料集能建立什麼?
- It guarantees the model works for every user and environment
- It validates performance across conditions not tested
- It explains the cause of every incorrect prediction
- It can illustrate classification, but not prove broad performance
- 哪一個專案是合適的低風險起點?
- Classifying nonpersonal recyclable materials
- Identifying students from face images without an approved need
- Recording classroom conversations as sound categories
- Predicting student achievement from personal characteristics
- What does an image classifier learn to output in a classroom project?
- A natural-language reason for each classification
- One of the categories represented by its labels
- A guaranteed description of an unfamiliar object
- A judgment about whether a category is socially appropriate
- How should a teacher describe on-device training?
- Proof that data can never leave the device in any workflow
- A reason account and export settings do not matter
- An available workflow choice, not a guarantee for every save or export path
- A guarantee that saved projects are private by default
- What should students do with false predictions?
- Change several parts of the dataset and guess which mattered
- Evaluate only examples used during training
- Remove every error from the report before sharing results
- Record and analyze them, then test a change