R
130 words and 439 other names, in italics, pointing to their entries.
- R² → coefficient of determination
- R1-Zero → DeepSeek-R1
- R2 → coefficient of determination
- race dynamics → AI arms race
- race to AGI → AI arms race
- rack-scale architecture → rack-scale system
- rack-scale computer → rack-scale system
- rack-scale system
- Rademacher complexity → statistical learning theory
- Radeon Open Compute → ROCm
- Alec Radford
- radial basis function → radial basis function network
- radial basis function kernel → kernel trick
- radial basis function network
- radiance field → neural radiance field
- radical behaviorism → behaviorism
- RAFT → optical flow
- RAG → retrieval-augmented generation
- RAI → responsible AI
- RAIL → model license
- Rainbow → deep Q-network
- RAISE Act
- Ralph → Ralph Wiggum loop
- Ralph loop → Ralph Wiggum loop
- Ralph Wiggum loop
- Ralph Wiggum technique → Ralph Wiggum loop
- ramification problem → frame problem
- ramp function → ReLU
- RAND → RAND Corporation
- RAND Corporation
- random → stochastic
- random decision forest → random forest
- random forest
- random forests → random forest
- random hyperparameter search → random search
- random projection → Johnson–Lindenstrauss lemma
- random sampling → sampling
- random search
- random seed
- random subspace method → random forest
- random variable
- random variate → random variable
- randomized probability matching → Thompson sampling
- randomized response → local differential privacy
- randomized search → random search
- randomized smoothing
- RandomizedSearchCV → random search
- randomness → stochastic
- random-restart hill climbing → hill climbing
- rank
- rank-deficient → rank
- rank–frequency law → Zipf's law
- RankNet → learning to rank
- Rank-One Model Editing → ROME
- rank-one model editing → ROME
- rapidly exploring random tree
- Rapture for nerds → rapture of the nerds
- rapture of the nerds
- RaR → rubric reward
- Rasch model → item response theory
- Rashomon effect
- Rashomon set → Rashomon effect
- rate limit
- rate limiting → rate limit
- rate of change → derivative
- rational agent → intelligent agent
- rational choice theory → decision theory
- rationalist community
- rationalist movement → rationalist community
- rationalist-adjacent → rationalist community
- rationalists → rationalist community
- ratsphere → rationalist community
- Ray
- Ray Data → Ray
- Ray J. Solomonoff → Ray Solomonoff
- Ray Serve → Ray
- Ray Train → Ray
- Ray Tune → Ray
- Raymond Kurzweil → Ray Kurzweil
- Raymond Solomonoff → Ray Solomonoff
- RBF kernel → kernel trick
- RBF network → radial basis function network
- RBFN → radial basis function network
- RBI → remote biometric identification
- RBM → restricted Boltzmann machine
- R-CNN
- RDF → Resource Description Framework
- RDF Schema → Resource Description Framework
- RDF triple → Resource Description Framework
- RDFS → Resource Description Framework
- RDMA
- RDMA over Converged Ethernet → RDMA
- reachability penalty → relative reachability
- ReAct
- ReAct agent → ReAct
- ReAct prompting → ReAct
- reactive control → subsumption architecture
- reading comprehension → question answering
- real drift → concept drift
- real log canonical threshold → singular learning theory
- reality gap → sim-to-real transfer
- RealNVP → normalizing flow
- real-time remote biometric identification → remote biometric identification
- real-to-sim → sim-to-real transfer
- reason → reasoning
- Reason + Act → ReAct
- reasonably foreseeable misuse → intended purpose
- reason-and-act → ReAct
- reasoner → reasoning model
- reasoning
- reasoning budget → thinking budget
- reasoning chain → reasoning trace
- reasoning effort → thinking budget
- reasoning engine → inference engine
- reasoning faithfulness → chain-of-thought faithfulness
- reasoning in latent space → latent reasoning
- reasoning LLM → reasoning model
- reasoning model
- reasoning monitoring → chain-of-thought monitoring
- reasoning tokens → reasoning trace
- reasoning trace
- RE-Bench
- REC → visual grounding
- recall
- recall at k → recall@k
- recall@k
- Recall-Oriented Understudy for Gisting Evaluation → ROUGE
- reCAPTCHA → CAPTCHA
- recaptioning → image captioning
- receding horizon control → model predictive control
- receiver operating characteristic → ROC curve
- Receptance Weighted Key Value → RWKV
- receptive field
- reciprocal rank → mean reciprocal rank
- reciprocal rank fusion
- recognizing textual entailment → natural language inference
- recommendation algorithm → recommender system
- recommendation engine → recommender system
- Recommendation on the Ethics of Artificial Intelligence → UNESCO Recommendation on the Ethics of AI
- recommendation system → recommender system
- recommender → recommender system
- recommender system
- reconstruction attack → model inversion
- record → example
- recourse → algorithmic recourse
- recsys → recommender system
- rectification → rectified flow
- rectified flow
- rectified flow transformer → rectified flow
- rectified linear unit → ReLU
- rectifier → ReLU
- recurrent network → recurrent neural network
- recurrent neural network
- recurrent processing theory
- recursive feature elimination → feature selection
- recursive language model
- recursive LM → recursive language model
- recursive reward modeling
- recursive self-improvement
- recursively self-improving AI → recursive self-improvement
- Red AI → Green AI
- red lines for AI → AI red lines
- red team → red teaming
- red teamer → red teaming
- red teaming
- red teaming with language models → automated red teaming
- Raj Reddy
- redpilled → -pilled
- red-team → red teaming
- reduced-impact AI → impact measure
- reduce-scatter → collective communication
- reducible loss → irreducible loss
- redundant encoding → proxy discrimination
- Redwood → Redwood Research
- Redwood Research
- ref model → reference model
- RefCOCO → visual grounding
- reference input → baseline
- reference model
- reference policy → reference model
- referring expression comprehension → visual grounding
- referring expression segmentation → visual grounding
- Reflexion
- Reflexion agent → Reflexion
- reflow → rectified flow
- Reformer → sparse attention
- ReFT → representation fine-tuning
- refusal
- refusal ablation → abliteration
- refusal behavior → refusal
- refusal direction
- refusal feature → refusal direction
- refusal training → refusal
- refusal vector → refusal direction
- refuse → refusal
- region proposal network → R-CNN
- Regions with CNN features → R-CNN
- ReGLU → SwiGLU
- regress → regression
- regression
- regression analysis → regression
- regression model → regression
- regression tree → decision tree
- regressor → regression
- regret
- regularisation → regularization
- regularization
- regularize → regularization
- regularized risk minimization → empirical risk minimization
- regularizer → regularization
- regulate AI → AI regulation
- Regulation (EU) 2016/679 → General Data Protection Regulation
- Regulation (EU) 2022/2065 → Digital Services Act
- Regulation (EU) 2024/1689 → EU AI Act
- Regulation (EU) 2026/1744 → Digital Omnibus
- regulation of AI → AI regulation
- regulatory sandbox → AI regulatory sandbox
- regurgitation → memorization
- REINFORCE
- REINFORCE leave-one-out
- REINFORCE with baseline → REINFORCE
- reinforced fine-tuning → reinforcement fine-tuning
- reinforcement fine-tuning
- reinforcement learning
- reinforcement learning environment → RL environment
- reinforcement learning from AI feedback
- reinforcement learning from human feedback
- reinforcement learning from human preferences → reinforcement learning from human feedback
- reinforcement learning from verifiable rewards → reinforcement learning with verifiable rewards
- reinforcement learning hypothesis → reward hypothesis
- reinforcement learning scaling → RL scaling
- reinforcement learning with verifiable rewards
- reject option → abstention
- rejection sampling
- rejection sampling fine-tuning → rejection sampling
- rejection-sampled fine-tuning → rejection sampling
- Re-LAION-5B → LAION-5B
- relation extraction → information extraction
- relative depth → depth estimation
- relative entropy → Kullback–Leibler divergence
- relative positional encoding → positional encoding
- relative reachability
- relevance propagation → layer-wise relevance propagation
- reliability diagram → calibration
- ReLU
- remat → gradient checkpointing
- rematerialization → gradient checkpointing
- Remotasks → Scale AI
- remote agent → background agent
- remote assistance → teleoperation
- remote attestation
- remote biometric identification
- remote biometric identification system → remote biometric identification
- remote direct memory access → RDMA
- Remote Labor Index
- remote operation → teleoperation
- remotely piloted aircraft system → drone
- removal of Sam Altman → OpenAI leadership crisis (2023)
- REP → robots.txt
- reparameterization gradient → reparameterization trick
- reparameterization trick
- reparametrization trick → reparameterization trick
- RepE → representation engineering
- repetition penalty
- replacement model
- replay buffer → experience replay
- replay memory → experience replay
- replicability → reproducibility
- representation control → representation engineering
- representation engineering
- representation fine-tuning
- representation finetuning → representation fine-tuning
- representation learning
- representation reading → representation engineering
- representation rerouting → circuit breakers
- representation steering → activation steering
- representational capacity → model capacity
- representational convergence → platonic representation hypothesis
- representational similarity analysis → centered kernel alignment
- reproducibility
- reproducibility checklist → reproducibility
- reproducibility crisis → reproducibility
- reproducible → reproducibility
- reproducible research → reproducibility
- reproducing kernel Hilbert space → kernel trick
- Reptile → model-agnostic meta-learning
- requests per minute → rate limit
- requests per second → throughput
- rerank → reranking
- re-ranker → reranking
- reranker → reranking
- reranking
- re-ranking → reranking
- resample ablation → ablation
- rescaling → feature scaling
- research agent → deep research
- Research Engineering Benchmark → RE-Bench
- Research Priorities for Robust and Beneficial Artificial Intelligence → Open Letter on Artificial Intelligence
- research taste
- reserved token → special token
- reservoir → reservoir computing
- reservoir computing
- residual block → residual connection
- residual connection
- residual learning → residual connection
- residual network → ResNet
- residual neural network → ResNet
- residual stream
- residual stream vector → residual stream
- ResNet
- ResNet-50 → ResNet
- ResNeXt → ResNet
- resolution
- resolution principle → resolution
- resolution refutation → resolution
- resolvent → resolution
- Resource Description Framework
- resource-poor language → low-resource language
- response time → latency
- responsibility gap
- responsibility gaps → responsibility gap
- responsible AI
- Responsible AI License → model license
- Responsible AI Safety and Education Act → RAISE Act
- responsible artificial intelligence → responsible AI
- responsible scaling policy
- restricted Boltzmann machine
- Rete algorithm → production system
- RetNet → linear attention
- retrieval-augmented generation
- retrieval-augmented LLM → retrieval-augmented generation
- return
- return-conditioned policy → decision transformer
- revenge porn → non-consensual intimate imagery
- reversal curse
- reversal paradox → Simpson's paradox
- reverse acquihire
- reverse acqui-hire → reverse acquihire
- reverse KL → Kullback–Leibler divergence
- reverse-mode differentiation → automatic differentiation
- revised Product Liability Directive → Product Liability Directive
- reward
- Reward Bench → RewardBench
- reward corruption → reward tampering
- reward exploitation → reward hacking
- reward function
- reward function tampering → reward tampering
- reward gaming → reward hacking
- reward hack → reward hacking
- reward hacking
- reward hypothesis
- reward input tampering → reward tampering
- reward is enough
- reward model
- reward model overoptimization → reward overoptimization
- reward modeling → reward model
- reward overoptimisation → reward overoptimization
- reward overoptimization
- reward over-optimization → reward overoptimization
- reward shaping
- reward signal → reward
- reward tampering
- RewardBench
- RewardBench 2 → RewardBench
- reward-free alignment → direct alignment algorithm
- reward-is-enough hypothesis → reward is enough
- reward-to-go → return
- RF → random forest
- RFT → reinforcement fine-tuning
- Rich Sutton → Richard Sutton
- Richard E. Bellman → Richard Bellman
- Richard Ernest Bellman → Richard Bellman
- Richard S. Sutton → Richard Sutton
- ridge → ridge regression
- ridge regression
- right of publicity
- right to an explanation → right to explanation
- right to explanation
- right to warn → AI whistleblower protections
- rights reservation → opt-out
- rigid-body simulator → physics simulator
- ring all-reduce → all-reduce
- ring attention → context parallelism
- risk from artificial intelligence → AI risk
- risk of astronomical suffering → suffering risk
- risk pyramid → risk-based approach
- risk-based approach
- risk-based regulation → risk-based approach
- risks from AI → AI risk
- risk-tiered approach → risk-based approach
- R@k → recall@k
- RKHS → kernel trick
- RL → reinforcement learning
- RL compute scaling → RL scaling
- RL env → RL environment
- RL environment
- RL fine-tuning → reinforcement fine-tuning
- RL for agents → agentic reinforcement learning
- RL gym → RL environment
- RL scaling
- RL scaling laws → RL scaling
- RLAIF → reinforcement learning from AI feedback
- RLCT → singular learning theory
- RLHF → reinforcement learning from human feedback
- RLI → Remote Labor Index
- RLlib → Ray
- RLM → recursive language model
- RLOO → REINFORCE leave-one-out
- RLVR → reinforcement learning with verifiable rewards
- RM → reward model
- RMS norm → RMSNorm
- RMS normalization → RMSNorm
- RMSE → mean squared error
- RMSNorm
- RMSProp
- RMU → WMDP
- RNG seed → random seed
- RNN → recurrent neural network
- roadmap planner → probabilistic roadmap
- RoBERTa → BERT
- robocar → autonomous vehicle
- Robodebt
- robo-debt → Robodebt
- Robodebt scheme → Robodebt
- robot
- robot arm → industrial robot
- robot dog → legged robot
- robot foundation model
- robot hand → end effector
- robot learning
- robot locomotion → locomotion
- robot manipulation → manipulation
- robot motion planning → motion planning
- Robot Operating System
- robot reply → Chinese room
- robot rights → AI personhood
- robot simulator → physics simulator
- robot skin → tactile sensing
- robot swarm → swarm robotics
- robot uprising → AI takeover
- robotaxi
- robo-taxi → robotaxi
- robotaxis → robotaxi
- robotic → robot
- robotic engineering → robotics
- robotic grasping → grasping
- robotic manipulation → manipulation
- robotic process automation
- robotics
- robotics foundation model → robot foundation model
- Robotics Transformer 2 → RT-2
- robots → robot
- Robots Exclusion Protocol → robots.txt
- robots exclusion standard → robots.txt
- robots.txt
- robust → robustness
- robust accuracy → adversarial robustness
- robust AI → robustness
- robust regression → Huber loss
- robust scaling → feature scaling
- robustness
- robustness certificate → certified robustness
- robustness to adversarial examples → adversarial robustness
- ROC → ROC curve
- ROC analysis → ROC curve
- ROC AUC → AUC
- ROC curve
- ROC space → ROC curve
- RoCE → RDMA
- ROCm
- Rocq
- Rocq Prover → Rocq
- Rodney A. Brooks → Rodney Brooks
- Rodney Allen Brooks → Rodney Brooks
- RoFormer → rotary positional embedding
- rogue agent → rogue AI
- rogue AI
- rogue AI agent → rogue AI
- rogue replication → autonomous replication and adaptation
- RoI pooling → R-CNN
- RoIAlign → R-CNN
- Roko's basilisk
- role prompting
- role-play framing → simulator theory
- role-play prompting → role prompting
- roll out → rollout
- rollback → model deployment
- rolling batching → continuous batching
- rollout
- roll-out → rollout
- ROME
- roofline → roofline model
- roofline analysis → roofline model
- roofline model
- roofline plot → roofline model
- root mean square layer normalization → RMSNorm
- root mean square propagation → RMSProp
- root mean squared error → mean squared error
- RoPE → rotary positional embedding
- ROS → Robot Operating System
- ROS 1 → Robot Operating System
- ROS 2 → Robot Operating System
- Frank Rosenblatt
- Rosenblatt perceptron → perceptron
- rotary embedding → rotary positional embedding
- rotary position embedding → rotary positional embedding
- rotary positional embedding
- rotation estimation → cross-validation
- ROUGE
- ROUGE score → ROUGE
- ROUGE-1 → ROUGE
- ROUGE-2 → ROUGE
- ROUGE-L → ROUGE
- ROUGE-N → ROUGE
- route planning → path planning
- router
- router z-loss → z-loss
- routing by agreement → capsule network
- rover → mobile robot
- row vector → vector
- Royal Commission into the Robodebt Scheme → Robodebt
- RPA → robotic process automation
- RPAS → drone
- RPM → rate limit
- RPN → R-CNN
- RPT → recurrent processing theory
- RRF → reciprocal rank fusion
- RRM → recursive reward modeling
- RRT → rapidly exploring random tree
- RRT* → rapidly exploring random tree
- RRT-Connect → rapidly exploring random tree
- r's in strawberry → strawberry problem
- RSA (neuroscience) → centered kernel alignment
- RSI → recursive self-improvement
- RSP → responsible scaling policy
- R-squared → coefficient of determination
- RSS → ICRA
- RT-1-X → Open X-Embodiment
- RT-2
- RT-2-X → RT-2
- RT-DETR → DETR
- RTE → natural language inference
- RT-X → Open X-Embodiment
- Rubin causal model → causal inference
- rubric grading → rubric reward
- rubric reward
- rubric-based reward → rubric reward
- rubrics as rewards → rubric reward
- rug pull (MCP) → tool poisoning
- rule engine → rule-based system
- rule of thumb → heuristic
- rule-based AI → rule-based system
- rule-based system
- rules engine → rule-based system
- rules file → agent instruction file
- David Rumelhart
- run → training run
- run tracking → experiment tracking
- runaway feedback loop → feedback loop
- running locally → on-device AI
- running out of data → data wall
- runtime retrieval → just-in-time retrieval
- Runway
- Runway ML → Runway
- RunwayML → Runway
- Stuart Russell
- Russellian monism → panpsychism
- RWKV
- RWKV-7 → RWKV