CS7643 last quiz Latest Update 2025-2026 50
Questions and 100% Verified Correct Answers
Guaranteed A+
Actor-Critic - CORRECT ANSWER: - Replaces rewards with Q_(PI_theta)(s, a)
- E[delta_theta * log_pi_theta(a | s) (Q_(PI_theta)(s, a))
Advantage Actor-critic - CORRECT ANSWER: - Uses Q minus V values (i.e. Advantage)
- E[delta_theta * log_pi_theta(a | s) (Q_(PI_theta)(s, a) - V_PI_theta)(s))
Approaches to Meta-Training - CORRECT ANSWER: 1. MatchingNet:
- Cosine distance of features between support and query set
2. ProtoNet
- Extract features from support and query set
- Take the mean of the features of the support set
- compare each query to the mean of the features (euclidean distance)
3. RelationNet
- Same as ProtoNet, but using a different distance function
- Relation Module learns how to relate in a more complicated manner than Cosine
Similarity or Euclidean Distance
Clustering Assumption and Deep Clustering - CORRECT ANSWER: - High density
regions forms a cluster while low density region separates clusters which hold a
coherent semantic meaning
, Avoid:
1. Empty Clusters
2. Trivial Parameterizations
Cons of Few-Shot Learning Baseline - CORRECT ANSWER: - The training does not
factor the task into account
--> No notion that we will be performing a bunch of N-way tests
Contrastive Loss - CORRECT ANSWER: Dot product between augmentation 1 and
positive & negative examples
Cosine Classifier - CORRECT ANSWER: - Cosine (similarity based) classifiers rather
than fully connected linear layers
- Effectively a dot product scaled to make a unit norm
--> only looking at the angles between feature vectors rather than their size
--> May provide better discrimination between small number of classes
Cross-View/Augmentation & Consistency - CORRECT ANSWER: - Take an unlabeled
example and make weakly and strongly augmented data
- Use weakly-augment an image and get a pseudo-label
- Strongly-augment an image and make a prediction
- train these predictions on the labels from the weakly augmented data
Idea:
- Weak augmentation isn't so severe that the pseudo-labels are bad
- Using strong augmentation to make the NN learn better feature representations
Deep Q-Learning - CORRECT ANSWER: - Q(s, a; w, b) = w_a^t * s + b_a
Questions and 100% Verified Correct Answers
Guaranteed A+
Actor-Critic - CORRECT ANSWER: - Replaces rewards with Q_(PI_theta)(s, a)
- E[delta_theta * log_pi_theta(a | s) (Q_(PI_theta)(s, a))
Advantage Actor-critic - CORRECT ANSWER: - Uses Q minus V values (i.e. Advantage)
- E[delta_theta * log_pi_theta(a | s) (Q_(PI_theta)(s, a) - V_PI_theta)(s))
Approaches to Meta-Training - CORRECT ANSWER: 1. MatchingNet:
- Cosine distance of features between support and query set
2. ProtoNet
- Extract features from support and query set
- Take the mean of the features of the support set
- compare each query to the mean of the features (euclidean distance)
3. RelationNet
- Same as ProtoNet, but using a different distance function
- Relation Module learns how to relate in a more complicated manner than Cosine
Similarity or Euclidean Distance
Clustering Assumption and Deep Clustering - CORRECT ANSWER: - High density
regions forms a cluster while low density region separates clusters which hold a
coherent semantic meaning
, Avoid:
1. Empty Clusters
2. Trivial Parameterizations
Cons of Few-Shot Learning Baseline - CORRECT ANSWER: - The training does not
factor the task into account
--> No notion that we will be performing a bunch of N-way tests
Contrastive Loss - CORRECT ANSWER: Dot product between augmentation 1 and
positive & negative examples
Cosine Classifier - CORRECT ANSWER: - Cosine (similarity based) classifiers rather
than fully connected linear layers
- Effectively a dot product scaled to make a unit norm
--> only looking at the angles between feature vectors rather than their size
--> May provide better discrimination between small number of classes
Cross-View/Augmentation & Consistency - CORRECT ANSWER: - Take an unlabeled
example and make weakly and strongly augmented data
- Use weakly-augment an image and get a pseudo-label
- Strongly-augment an image and make a prediction
- train these predictions on the labels from the weakly augmented data
Idea:
- Weak augmentation isn't so severe that the pseudo-labels are bad
- Using strong augmentation to make the NN learn better feature representations
Deep Q-Learning - CORRECT ANSWER: - Q(s, a; w, b) = w_a^t * s + b_a
