D (35 terms)

Data Augmentation Expanding training data via transformations (flips, noise, paraphrases) to improve robustness. Intermediate Data Drift Shift in feature distribution over time. Intermediate Data Governance Processes and controls for data quality, access, lineage, retention, and compliance across the AI lifecycle. Intermediate Data Labeling Human or automated process of assigning targets; quality, consistency, and guidelines matter heavily. Intermediate Data Leakage When information from evaluation data improperly influences training, inflating reported performance. Intermediate Data Lineage Tracking where data came from and how it was transformed; key for debugging and compliance. Intermediate Data Poisoning Maliciously inserting or altering training data to implant backdoors or degrade performance. Intermediate Data Protection Impact Assessment Privacy risk analysis under GDPR-like laws. Intermediate Data Scaling Increasing performance via more data. Intermediate Dataset A structured collection of examples used to train/evaluate models; quality, bias, and coverage often dominate outcomes. Intermediate Dataset Shift Differences between training and deployed patient populations. Intermediate Datasheet for Datasets Structured dataset documentation covering collection, composition, recommended uses, biases, and maintenance. Intermediate Deceptive Alignment Model behaves well during training but not deployment. Advanced Decomposition Prompt Breaking tasks into sub-steps. Intro Deep Learning A branch of ML using multi-layer neural networks to learn hierarchical representations, often excelling in vision, sp... Intermediate Deliberative Agent Agent reasoning about future outcomes. Advanced Delimited Prompt Using markers to isolate context segments. Intro Denoising Diffusion Probabilistic Model Diffusion model trained to remove noise step by step. Advanced Depth vs Width Tradeoffs between many layers vs many neurons per layer. Intermediate Developmental Robotics Robots learning via exploration and growth. Advanced Differential Privacy A formal privacy framework ensuring outputs do not reveal much about any single individual’s data contribution. Intermediate Differential Progress Accelerating safety relative to capabilities. Intermediate Diffusion Model Generative model that learns to reverse a gradual noise process. Advanced Digital Twin High-fidelity virtual model of a physical system. Advanced Distillation Training a smaller “student” model to mimic a larger “teacher,” often improving efficiency while retaining performance. Intermediate Distribution Shift Train/test environment mismatch. Intermediate Do-Operator Models effects of interventions (do(X=x)). Advanced Domain Randomization Randomizing simulation parameters to improve real-world transfer. Advanced Domain Shift A mismatch between training and deployment data distributions that can degrade model performance. Intermediate DPO A preference-based training method optimizing policies directly from pairwise comparisons without explicit RL loops. Intermediate Dropout Randomly zeroing activations during training to reduce co-adaptation and overfitting. Intermediate Dual Problem Alternative formulation providing bounds. Intermediate Due Process Legal right to fair treatment. Intermediate Dynamics Motion considering forces and mass. Advanced Dynamics Model Predicts next state given current state and action. Advanced