引導式測驗 · 中等 等級
馬可夫鏈測驗
測試馬可夫性質、轉移矩陣行、兩步驟天氣路徑、平穩分佈、狀態設計和簡單鏈的限制。
問題 1 的 8
In this guide, what does the Markov property say about predicting the next state?
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。
- In this guide, what does the Markov property say about predicting the next state?
- Only the first state matters, regardless of the current one
- The next state is fixed once the current state is named
- Given the current state, earlier states add no information under the model
- The process must have no hidden real-world history
- Why must each row of the guide's transition matrix sum to one?
- A row gives probabilities for every prior state
- A row contains only the stationary probabilities
- Every entry in a probability matrix must be identical
- It lists all mutually exclusive next-state probabilities from one current state
- If today is sunny in the guide's illustrative matrix, what is the probability of rain tomorrow?
- 0.8
- 0.2
- 0.4
- 0.6
- Starting sunny, what is the guide's illustrative probability of rain two days later?
- 0.20, using only the direct one-step route
- 0.60, using only the rainy-state row
- 0.80, using the sunny-to-sunny probability
- 0.28, adding paths through sunny and rainy tomorrow
- What makes a state distribution stationary for a transition matrix?
- It changes as quickly as possible after each step
- Applying one more transition leaves that distribution unchanged
- It describes only the initial state before any transitions
- It assigns equal probability to every state in every chain
- Which sunny/rainy mixture is stationary for the guide's illustrative matrix?
- One-third sunny and two-thirds rainy
- One-half sunny and one-half rainy
- Two-thirds sunny and one-third rainy
- All sunny and no rainy days
- A machine's failure risk depends on how long it has been running, but the state records only working or broken. What is the concern?
- The chosen state may omit history needed to model the next transition
- Every transition probability must become zero
- The stationary distribution is automatically uniform
- Two states make transition matrices impossible
- Why can a simple short-state Markov model struggle with longer passages of text?
- It may miss long-range language dependencies outside its state
- A fixed transition matrix makes every next token equally likely
- It cannot assign a next-token probability unless the entire text is stored
- More training observations automatically add the missing earlier context to the state