Machine Learning Trivia Questions
- According to Mercer’s Theorem, what property must a kernel function's Gram matrix possess to be valid for use in an SVM?
- In a Convolutional Neural Network (CNN), what term describes the specific region in the input space that a particular feature is looking at?
- In a Convolutional Neural Network, what is the name of the parameter that determines the number of pixels the filter moves across the input matrix?
- In a Generative Adversarial Network (GAN), which component is trained to identify whether an image is real or synthetic?
- In contrast to standard Autoencoders, Variational Autoencoders (VAEs) map input data to what kind of latent space to allow for generative sampling?
- In Gradient Boosting, what type of model is typically used as the 'weak learner'?
- In imitation learning, a behavioral-cloning policy with small per-step error can accumulate total error growing quadratically with the horizon T; DAgger's no-regret formulation reduces this growth to what order in T?
- In mechanistic interpretability, what type of model is trained on a network's internal activations to decompose them into a large set of sparse, more human-interpretable features, helping address superposition?
- In Principal Component Analysis (PCA), the principal components are the eigenvectors of which specific matrix?
- In Reinforcement Learning, what equation expresses the value of a state as the expected immediate reward plus the discounted value of the next state?
- In Reinforcement Learning, what is the term for the tension between trying new actions to find better rewards and choosing known actions that yield high rewards?
- In Support Vector Machines, what is the 'Kernel Trick' used to achieve?
- In the Bias-Variance Tradeoff, what component of error is reduced by increasing model complexity?
- In the bias-variance tradeoff, what usually happens to the variance as the model complexity (degrees of freedom) increases?
- In the context of bias-variance tradeoff, does a high-bias model typically underfit or overfit the data?
- In the context of dimensionality reduction, PCA identifies directions that maximize what statistical property?
- In the context of ensemble learning, what is the term for sampling the training data with replacement to create multiple diverse models?
- In the context of LSTMs, which gate decides what information from the previous cell state should be discarded?
- In the context of Recurrent Neural Networks (RNNs), what phenomenon occurs when gradients converge to zero, making it difficult to learn long-range dependencies?
- In the ID3 decision tree algorithm, which metric (calculated using entropy) is used to select the attribute that best separates the training examples?
- In the optimization of Support Vector Machines, what set of necessary and sufficient conditions must be met for a solution to be optimal?
- In the standard RLHF pipeline, what intermediate model is trained to predict human preference rankings so that it can supply a learned reward signal during reinforcement learning?
- In the Transformer architecture, what mechanism is used to provide the model with information about the order of symbols in a sequence?
- Is Q-learning categorized as an on-policy or off-policy reinforcement learning algorithm?
- K-means clustering aims to minimize which specific metric, also known as the within-cluster sum-of-squares (WCSS)?
- The 'Vanishing Gradient Problem' is most commonly associated with which types of activation functions in deep networks?
- What 2013 technique by Mikolov et al. uses a shallow neural network to create dense vector representations of words?
- What abbreviation names the alignment technique that trains a reward model on human preference comparisons, then fine-tunes a language model against it with reinforcement learning?
- What are the three components of the 'Self-Attention' mechanism's calculation?
- What breakthrough 2017 paper introduced the Transformer architecture, which is built around self-attention?
- What core mechanism allows Transformer models to process all parts of an input sequence simultaneously while weighing their relative importance?
- What does the optimization algorithm name 'Adam' stand for?
- What does the 'PAC' stand for in Leslie Valiant's mathematical framework for analyzing the complexity of learning algorithms?
- What evaluation metric is defined as the harmonic mean of precision and recall?
- What general term describes the failure mode in which a model's deployment input distribution differs from its training distribution, degrading performance, and which sequential imitation learners induce on themselves?
- What heuristic is commonly used in K-means clustering to determine the optimal number of clusters by looking for a 'kink' in the cost function?
- What is the common name for the regularization technique that adds the squared magnitude of coefficients as a penalty term to the loss function?
- What is the computational complexity of the self-attention mechanism in a standard Transformer relative to the sequence length L?
- Which SVM kernel is defined by K(x, x′) = exp(−γ‖x − x′‖²)?
- What is the name for the empirical power-law relationships showing that a language model's loss falls predictably as model size, dataset size, and training compute increase?
- What is the name of the alignment method that replaces many human preference labels with model self-critique and revision guided by a written set of principles?
- What is the name of the problem of inferring the reward function an expert is optimizing purely from observations of their behavior, rather than being given the reward directly?
- What is the name of the theorem stating that a feed-forward network with a single hidden layer and a nonpolynomial activation function can approximate any continuous function on a compact set to arbitrary accuracy?
- What is the name of the value function that estimates the total reward an agent can expect to collect from a given state-action pair in Reinforcement Learning?
- What is the standard name for the loss function usually optimized in logistic regression for binary classification?
- What is the term for a language model's ability to perform a new task purely from instructions and examples provided in its prompt, without any updates to its weights?
- What is the term for the learning setting in which the algorithm itself chooses which unlabeled examples to have labeled, typically the most uncertain or informative ones, to reach high accuracy with fewer labels?
- What is the term for the simplest form of imitation learning, in which a policy is trained with supervised learning to directly map states to an expert's demonstrated actions?
- What is the term for the value added to the denominator in Batch Normalization to prevent division by zero?
- What is the time complexity of the standard self-attention mechanism with respect to the sequence length n?