C (65 terms)

Caching Storing results to reduce compute. Intermediate Calibration The degree to which predicted probabilities match true frequencies (e.g., 0.8 means ~80% correct). Intermediate Canary Release Incrementally deploying new models to reduce risk. Intermediate Canary Tokens Detecting unauthorized model outputs or data leaks. Intermediate Capability Overhang Stored compute or algorithms enabling rapid jumps. Advanced Case Outcome Prediction Predicting case success probabilities. Intermediate Catastrophic Forgetting Loss of old knowledge when learning new tasks. Intermediate Causal Graph Directed acyclic graph encoding causal relationships. Advanced Causal Inference Framework for reasoning about cause-effect relationships beyond correlation, often using structural assumptions and e... Intermediate Causal Mask Prevents attention to future tokens during training/inference. Intermediate Central Limit Theorem Sum of independent variables converges to normal distribution. Advanced Chain-of-Thought Stepwise reasoning patterns that can improve multi-step tasks; often handled implicitly or summarized for safety/priv... Intermediate Change Management Governance of model changes. Intermediate Change Point Detection Identifying abrupt changes in data generation. Intermediate Chinchilla Scaling Scaling law optimizing compute vs data. Intermediate Chunking Breaking documents into pieces for retrieval; chunk size/overlap strongly affect RAG quality. Intermediate CI/CD for ML Automated testing and deployment processes for models and data workflows, extending DevOps to ML artifacts. Intermediate Class Imbalance When some classes are rare, requiring reweighting, resampling, or specialized metrics. Intermediate Clinical Decision Support AI systems assisting clinicians with diagnosis or treatment decisions. Intermediate Clinical Validation Testing AI under actual clinical conditions. Intermediate CLIP Joint vision-language model aligning images and text. Intermediate Closed Model Models accessible only via service APIs. Intermediate Closed-Loop Control Control using real-time sensor feedback. Advanced Cognitive Architecture System-level design for general intelligence. Frontier Cognitive Map Internal representation of environment layout. Frontier Cold Start Startup latency for services. Intermediate Commoditization Declining differentiation among models. Intermediate Commonsense Physics Human-like understanding of physical behavior. Frontier Competitive Game Agents have opposing objectives. Advanced Computational Chemistry Modeling chemical systems computationally. Advanced Computational Learning Theory A theoretical framework analyzing what classes of functions can be learned, how efficiently, and with what guarantees. Intermediate Compute Hardware resources used for training/inference; constrained by memory bandwidth, FLOPs, and parallelism. Intermediate Compute Governance Regulating access to large-scale compute. Intermediate Compute Scaling Increasing model capacity via compute. Intermediate Computer Vision AI focused on interpreting images/video: classification, detection, segmentation, tracking, and 3D understanding. Intermediate Computer-Aided Diagnosis Automated assistance identifying disease indicators. Intermediate Concept Drift The relationship between inputs and outputs changes over time, requiring monitoring and model updates. Intermediate Condition Number Sensitivity of a function to input perturbations. Advanced Conditional Random Field Probabilistic graphical model for structured prediction. Intermediate Configuration Space Space of all possible robot configurations. Advanced Confounding A hidden variable influences both cause and effect, biasing naive estimates of causal impact. Intermediate Confusion Matrix A table summarizing classification outcomes, foundational for metrics like precision, recall, specificity. Intermediate Constrained Optimization Optimization under equality/inequality constraints. Intermediate Constraint Prompting Explicit output constraints (format, tone). Intro Contact Dynamics Modeling interactions with environment. Advanced Context Compression Techniques to handle longer documents without quadratic cost. Intermediate Context Window Maximum number of tokens the model can attend to in one forward pass; constrains long-document reasoning. Intermediate Control Loop Continuous loop adjusting actions based on state feedback. Advanced Control Theory Mathematical framework for controlling dynamic systems. Intermediate Controller Algorithm computing control actions. Intermediate Convex Optimization Optimization problems where any local minimum is global. Intermediate Convolutional Neural Network Networks using convolution operations with weight sharing and locality, effective for images and signals. Intermediate Cooperative AI Designing AI to cooperate with humans and each other. Intermediate Cooperative Game Agents optimize collective outcomes. Advanced Coordination Failure Agents fail to coordinate optimally. Advanced Correlation Normalized covariance. Advanced Corrigibility Willingness of system to accept correction or shutdown. Advanced Cost Attribution Assigning AI costs to business units. Intermediate Counterfactual What would have happened under different conditions. Advanced Covariance Measures joint variability between variables. Advanced Credit Scoring Predicting borrower default risk. Intermediate Cross-Attention Attention between different modalities. Intermediate Cross-Entropy Measures divergence between true and predicted probability distributions. Intermediate Cross-Validation A robust evaluation technique that trains/evaluates across multiple splits to estimate performance variability. Intermediate Curriculum Learning Ordering training samples from easier to harder to improve convergence or generalization. Intermediate