The Curious Engineer

All blogs / AI / ML

Apple Machine Learning Research

machinelearning.apple.com

Apple machine learning teams are engaged in state of the art research in machine learning and artificial intelligence. Learn about the latest advancements.

AI / ML

’21’22’23’24’25’26
Posts
1,240 from 24 Jun 2021 to 26 Oct 2026
Posting
weekly (94 in the last year)
Kind
company blog
Feed
https://machinelearning.apple.com/rss.xml

2026 84

  1. GHARP: Real-Time Gaussian Head Animation from Large-Scale Reconstruction Prior
  2. Empirical Methods in Natural Language Processing (EMNLP) 2026
  3. Local and Distributed Large LLM Inference with MLX and RDMA on Mac Studio
  4. Selection-Based Structured Reasoning: Toward Efficient Multimodal Search Agents
  5. Plan or Retrieve First? A Scale Study of Tool Discovery Architectures
  6. Agent Seer: Synthesizing Scenarios from Specification Understanding
  7. Code-Switching Detection via Language-Selective Dimensions in Speech Foundation Models
  8. The Design Space of Tri-Modal Masked Diffusion Models
  9. Progressive Refinement: An Iterative Pseudo-Labeling Approach for Mandarin-English Code-Switching ASR
  10. Mix, Don't Tune: Bilingual Pre-Training Outperforms Hyperparameter Search in Data-Constrained Settings
  11. Large Language Models for Machine Translation Quality Annotation: Humans and Models Are Both Challenged
  12. Minions: A Dynamic and Distributed Virtual GPU Cloud Built on Idle Apple Silicon Devices
  13. Self-Compacting Language Model Agents
  14. Examining Human-Like Behaviors in LLMs: A Multi-Dimensional Analysis of Model Behaviors, User Factors, and System Prompts
  15. Normalizing Trajectory Models
  16. From Preferences to Principles: Rubric-Based Alignment for Grounded Knowledge Answers
  17. Anti-Causal Domain Generalization: Leveraging Unlabeled Data
  18. Behavioral Privacy Leakage in Agentic Negotiation: Formalizing and Mitigating Inference Attacks via Randomized Policies
  19. SCLATE: A Substrate for Continual-Learning Agent Training and Evaluation
  20. How Much of a Harness Does a Strong Agent Need for Autonomous ML Engineering?
  21. Shared Selective Persistent Memory for Agentic LLM Systems
  22. Unmasking On-Policy Distillation: Where It Helps, Where It Hurts, and Why
  23. On the Effectiveness-Fluency Trade-Off in LLM Conditioning: A Systematic Study
  24. Sparse Autoencoders Are Capable LLM Jailbreak Mitigators
  25. LinEAS: End-to-end Learning of Activation Steering with a Distributional Loss
  26. ExpertLens: Activation Steering Features Are Highly Interpretable
  27. Taming Text-to-Sounding Video Generation via Advanced Modality Condition and Interaction
  28. BalCapRL: A Balanced Framework for RL-Based MLLM Image Captioning
  29. research area Computer Vision
  30. PROOF-Gen: From Optimized Data to Better Distillation
  31. RISED: Rubrics for Agentic Multi-Environment Selection and Self-Distillation
  32. Nine Judges, Two Effective Votes: Correlated Errors Undermine LLM Evaluation Panels
  33. mAceReason-Math: A Dataset of High-Quality Multilingual Math Problems Ready For RLVR
  34. DACA-GRPO: Denoising-Aware Credit Assignment for Reinforcement Learning in Diffusion Language Models
  35. When Unlearning Is Free: Leveraging Low Influence Points to Reduce Computational Costs
  36. Trace Length is a Simple Uncertainty Signal in Reasoning Models
  37. When Unlearning Is Free: Leveraging Low Influence Points to Reduce Computational Costs
  38. Doubly Sub-linear Interactive Proofs of Proximity
  39. Interactive Proofs for General Distribution Properties
  40. Location-Invariant Properties of Functions Versus Properties of Distributions: United in Testing but Separated in Verification
  41. Negotiating Ontological Boundaries in User-Authored Personal Sensing Systems
  42. Limits of Confidence in Diffusion
  43. Compressing Streaming Neural Audio Encoders via Latent-Space Distillation
  44. LVSum: A Benchmark for Timestamp-Aware Long Video Summarization
  45. NarrativeTrack: Evaluating Entity-Centric Reasoning for Narrative Understanding
  46. RayRoPE: Projective Ray Positional Encoding for Multi-View Attention
  47. Annotations Mitigate Post-Training Mode Collapse
  48. Accelerating Text-to-Video Generation with Calibrated Sparse Attention
  49. Drop-In Perceptual Optimization for 3D Gaussian Splatting
  50. Residual Context Diffusion Language Models
  51. Learning Unmasking Policies for Diffusion Language Models
  52. Learning Deformable Body Interactions With Adaptive Spatial Tokenization
  53. Introducing the Third Generation of Apple’s Foundation Models
  54. Memory Efficient Audio Synthesis with Decoupled Temporal Depth Diffusion Transformers
  55. SpeakStream: Streaming Text-to-Speech with Interleaved Data
  56. Learning to Evict from Key-Value Cache
  57. Putting Captions to the Test: Evaluating Video Caption Quality through Multiple-Choice Question Answering
  58. Luce: Relightable Gaussians for 3D Asset Generation
  59. MoEs Are Stronger than You Think: Hyper-Parallel Inference Scaling with RoE
  60. EpiCache: Episodic KV Cache Management for Long-Term Conversation on Resource-Constrained Environments
  61. DiscoSign: Discourse-Aware Text to Sign Language Gloss Translation
  62. STARFlow2: Bridging Language Models and Normalizing Flows for Unified Multimodal Generation
  63. Conformal Thinking: Risk Control for Reasoning on a Compute Budget
  64. Local Mechanisms of Compositional Generalization in Conditional Diffusion
  65. ParaRNN: Large-Scale Nonlinear RNNs, Trainable in Parallel
  66. Scaling Properties of Continuous Diffusion Spoken Language Models
  67. GRPO Beyond English: A Large-Scale Study of GRPO in Non-English and Multilingual Settings
  68. MemoryLLM: Plug-n-Play Interpretable Feed-Forward Memory for Transformers
  69. Amortizing Maximum Inner Product Search with Learned Support Functions
  70. Understanding Annotator Safety Policy with Interpretability
  71. VideoFlexTok: Flexible-Length Coarse-to-Fine Video Tokenization
  72. Training Software Engineering Agents and Verifiers with SWE-Gym
  73. CPEP: Contrastive Pose-EMG Pre-training Enhances Gesture Generalization on EMG Signals
  74. Beyond Visual CoT: Internalized Visual Thinking for Proactive Video Reasoning
  75. On Robustness and Chain-of-Thought Consistency of RL-Finetuned VLMs
  76. Faster Rates for Federated Variational Inequalities
  77. A Single Neuron Is Sufficient to Bypass Safety Alignment in Large Language Models
  78. Multilingual Semantic Retrieval for Apple Music Search
  79. REVERSAL-BENCH: A Reversibility Axis and Reset Oracle for Measuring the Reset-Free RL Cliff
  80. Multi-Agent Teams Hold Experts Back
  81. COMPASS: Benchmarking Constrained Optimization in LLM Agents
  82. AMUSE: Audio-Visual Benchmark and Alignment Framework for Agentic Multi-Speaker Understanding
  83. Incentivizing Temporal-Awareness in Egocentric Video Understanding Models
  84. Fortress: A Case Study in Stabilizing Search Recommendations via Temporal Data Augmentation and Feature Pruning

2025 52

  1. Scaling Categorical Flow Maps
  2. MT-EditFlow: Reinforcement Learning for Multi-Turn Image Editing with Flow Matching
  3. Segmental Attention Decoding with Long Form Acoustic Encodings
  4. Arbitrage: Efficient Reasoning via Advantage-Aware Speculation
  5. Taming Outlier Tokens in Diffusion Transformers
  6. Glyph: A Multi-Strategy Agentic System for Column Description and Sensitivity-Ontology Tagging of Enterprise Data Catalogs
  7. Using LLMs for Late Multimodal Sensor Fusion for Activity Recognition
  8. In a previous post
  9. Dynamically Scaled Activation Steering
  10. One Layer Is Enough: Adapting Pretrained Visual Encoders for Image Generation
  11. Visatronic: A Multimodal Decoder-Only Model for Speech Synthesis
  12. LLMs Are Not (Consistently) Bayesian: Quantifying Internal (In)consistencies of LLMs’ Probabilistic Beliefs
  13. How to Guide Your Language Flow
  14. How Value Induction Reshapes LLM Behaviour
  15. Trajectory as the Teacher: Few-Step Discrete Flow Matching via Energy-Navigated Distillation
  16. Local Mechanisms of Compositional Generalization in Conditional Diffusion
  17. Dimensionality Reduction Meets Network Science: Sensemaking on UMAP’s kNN Graph
  18. Scaling Laws for Mixture Pretraining Under Data Constraints
  19. Language Discrimination Improves Linguistic Learning in Multilingual Speech Models
  20. Assessing the Role of Data Quality in Training Bilingual Language Models
  21. SimpleDesign: A Joint Model for Protein Sequence and Structure Codesign
  22. MVICAD2: Multi-View Independent Component Analysis with Delays and Dilations
  23. Understanding Alignment in Multimodal LLMs: A Comprehensive Study
  24. Removing Noise, not Finding Gold: Quality Filtering for Large-Scale Pretraining
  25. Emergent Symbolic Structure in Health Foundation Models: Extraction, Alignment, and Cross-Modal Transfer
  26. Recursive Language Models Meet Uncertainty: The Surprising Effectiveness of Self-Reflective Program Search for Long Context
  27. Beyond Next-Token Prediction: A Performance Characterization of Diffusion versus Autoregressive Language Models
  28. STARFlow-V: End-to-End Video Generative Modeling with Normalizing Flows
  29. VSAS-Bench: Real-Time Evaluation of Visual Streaming Assistant Models
  30. FlowEval: Reference-Based Evaluation of Generated User Interfaces
  31. Path-Constrained Mixture-of-Experts
  32. Uncertainty Quantification for LLM Function-Calling
  33. DynaMiCS: Fine-Tuning LLMs with Performance Constraints Using Dynamic Mixtures
  34. Efficient Privacy Loss Accounting for Subsampling and Random Allocation
  35. The Communication Bottleneck: A Round-Trip Study of Tree-Structured Expression Serialization in Language Models
  36. Updates to Apple’s On-Device and Server Foundation Language Models
  37. Learning Structured Reasoning via Tractable Trajectory Control
  38. MoMo: Dial Motion Mode in Robot Manipulation with Spatiotemporal Action Tokenization
  39. Rooms from Motion: Un-posed Indoor 3D Object Detection as Localization and Mapping
  40. Learning from Self Critique and Refinement for Faithful LLM Summarization
  41. Normalized Rewards for Preference Learning
  42. Understanding Aggregate Trends for Apple Intelligence Using Differential Privacy
  43. Length Value Model: Scalable Value Pretraining for Token-Level Length Modeling
  44. Optimal Splitting of Language Models from Mixtures to Specialized Domains
  45. Flexible Routing via Uncertainty Decomposition
  46. RLTL;DR: Self-Improvement by Internalizing Self-Generated Feedback
  47. REFACTOR-VLA: Unsupervised Library Learning of Typed Motor Programs
  48. (1D) Ordered Tokens Enable Efficient Test-Time Search
  49. Environment-free Synthetic Data Generation for API-Calling Agents
  50. Proactive Agent Research Environment: Simulating Active Users to Evaluate Proactive Assistants
  51. Revisiting ASR Error Correction with Specialized Models
  52. Large-Scale High-Quality 3D Gaussian Head Reconstruction from Multi-View Captures

2024 18

  1. Personalizing Incremental Video Search with Hybrid Text and ID Embeddings
  2. TopoPrimer: The Missing Topological Context in Forecasting Models
  3. Apple Machine Learning Research at NeurIPS 2024
  4. MODUS: Decoder-only Any-to-Any Modeling of Diverse Modalities
  5. LEAD: Breaking the No-Recovery Bottleneck in Long-Horizon Reasoning
  6. The P-Completeness of Inverted Index Traversal: On the Complexity of Evaluating Boolean Query DAGs
  7. RVPO: Risk-Sensitive Alignment via Variance Regularization
  8. CLaRa: Bridging Retrieval and Generation with Continuous Latent Reasoning
  9. LensVLM: Selective Context Expansion for Compressed Visual Representation of Text
  10. Weblica: Scalable and Reproducible Training Environments for Visual Web Agents
  11. Embarrassingly Simple Self-Distillation Improves Code Generation
  12. Athena: Intermediate Representations for Iterative Scaffolded App Generation with an LLM
  13. Sample and Map from a Single Convex Potential: Generation using Conjugate Moment Measures
  14. Locking Pretrained Weights via Deep Low-Rank Residual Distillation
  15. Balanced LoRA: Removing Parameter Invariance to Accelerate Convergence
  16. Introducing Apple’s On-Device and Server Foundation Models
  17. Device Passport: Enabling Spatio-Temporal Pretrained Models to Generalize Across Input Layouts
  18. A Specialized Semismooth Newton Method for Kernel-Based Optimal Transport

2023 6

  1. DarkDiff: Advancing Low-Light Raw Enhancement by Retasking Diffusion Models for Camera ISP
  2. IDEA Prune: An Integrated Enlarge-and-Prune Pipeline in Generative Language Model Pretraining
  3. Multilingual Knowledge Transfer under Data Constraints via Lexical Interventions
  4. ODKE+: Ontology-Guided Open-Domain Knowledge Extraction with LLMs
  5. A Practical Recipe for Semi-Supervised Federated ASR: Online Pseudo-Labels with Server Update Stabilization
  6. What Matters in Practical Learned Image Compression

2022 2

  1. Metric-Dependent Annotation Saturation for Learning from Label Distributions
  2. DeepAmbigQA: Ambiguous Multi-hop Questions for Benchmarking LLM Answer Completeness

2021 3

  1. Show Me Examples: Inferring Visual Concepts from Image Sets
  2. Velox: Learning Representations of 4D Geometry and Appearance
  3. GH-ESD: Grounded Hypothesis-Driven Error Slice Discovery for Instance-Level Vision Tasks

Undated 1075

  1. Text conditional jepa visual representations
  2. Specmd expert prefetching
  3. Normalizing flows iterative denoising
  4. Spatial
  5. Stochastic kv routing
  6. Portool policy optimization
  7. Simplefold
  8. Cram less
  9. Evaluating sample utility
  10. Reinforced agent inference feedback
  11. Stereofoley object aware stereo audio
  12. Sign language annotations
  13. Direct steering optimization
  14. Iclr 2026
  15. Adaptive thinking
  16. Ladir
  17. Bed llm
  18. Depyf open
  19. Long term motion embeddings
  20. Unigen 1.5
  21. Pico banana
  22. So bench
  23. Atoken
  24. Llm context understanding
  25. What do your logits know
  26. Mixatlas
  27. Mobileclip2
  28. Semidiscrete couplings
  29. Lacy
  30. Neighbor
  31. Governance aware agent telemetry
  32. Squire
  33. Personalized group
  34. Protext gender bias benchmark
  35. Depth pro
  36. Exclusive self attention
  37. Cognition emerge frontier models
  38. Entropy preserving reinforcement learning
  39. Beyond real data
  40. Less gaussians texture more
  41. To infinity
  42. Downstream metrics
  43. Fs dfm
  44. Latent lookahead
  45. Scaling synthetic task
  46. Safetypairs
  47. Trained on tokens
  48. Goldilocks
  49. Prose2policy
  50. Trajtok
  51. Ames
  52. Rubicap
  53. Multilingual reasoning gym
  54. Lito
  55. Augmenting app
  56. Multi frequency fusion
  57. Genctrl
  58. Completed hyperparameter
  59. Hallucination span detection
  60. Disentangled safety
  61. Embridge
  62. Separating intelligence
  63. We notice
  64. Abstral
  65. Diffucoder
  66. Hilbert
  67. Compute optimal
  68. Compute multiplier
  69. Constructive circuit amplification
  70. Automated recycling identification system
  71. Beyond a single extractor
  72. Closing the gap
  73. Cot
  74. Simulation based inference
  75. Query auto completion
  76. Ferret ui
  77. Correctness
  78. Semantic caching
  79. Scenescout
  80. Leveraging audio visual
  81. Data centric lessons
  82. Controlled experimentation
  83. Faster rates
  84. Mapping
  85. Pretrained foundation model
  86. Score distillation
  87. Egodex learning dexterous manipulation
  88. Parallel track
  89. Fairly certain
  90. Parts assemble
  91. Vssflow
  92. Bootstrapped
  93. Pararnn
  94. Continuously augmented
  95. Efficient calibration
  96. Reinforcement learning
  97. Gaussian process
  98. Sharp monocular view
  99. Manzano
  100. Grace
  101. Self supervised representations
  102. Designer feedback
  103. Vlsu
  104. Action abstractions
  105. Selfreflect
  106. Barriers for learning
  107. Hierarchical memories
  108. Rethinking jepa
  109. Rl for reasoning
  110. Semantic regexes
  111. Coarse grained
  112. Neurips 2025
  113. Search augmented
  114. Optimal transport
  115. https://machinelearning.apple.com/research/r1
  116. Agentbuilder
  117. Best of n
  118. Taxonomy for speech
  119. Inferring optical tissue
  120. Fairness dynamics
  121. Communication complexity
  122. Agentic reasoning
  123. Unified open
  124. Gie bench
  125. Inflectional morphology probes
  126. Sinkhorn couplings
  127. Mirror
  128. Moe phds
  129. Dit air
  130. Chipchat
  131. Digital mental health
  132. Algorithms
  133. Chain of sketch
  134. Reinforcement learning integrated
  135. Semantic mastery
  136. Predict
  137. Photoplethysmography
  138. Immersive debugging
  139. Vlsu mapping
  140. Speech foundation
  141. Relative composition eeg
  142. Omni router
  143. Illusion of thinking
  144. Car flow
  145. Starflow
  146. Polynorm
  147. Rattention
  148. Optimal data mixtures
  149. Embedding atlas
  150. Policy maps
  151. Semorec
  152. Reasoning razor
  153. Toward machine interpreting
  154. Leveraging power
  155. Psychological scaffolds
  156. Hallucination detection
  157. Memory efficient backpropagation
  158. Primex
  159. Simulation based
  160. Breaking down
  161. Persistent discrimination
  162. Switchboard affect
  163. Rethinking non negative
  164. Software defect prediction
  165. Hybrid vector graph
  166. Encqa benchmarking
  167. Preamble
  168. Clip up
  169. Discriminating form
  170. Llm inference
  171. Analyzing dialectical
  172. Reasoning based anomaly
  173. Stable diffusion
  174. Keyframer
  175. Taser
  176. Graph representation
  177. Proactive streaming assistant
  178. Privacy amplification
  179. Enabling
  180. Checklists are better
  181. Ai red teaming
  182. Prediction potential
  183. Adversarial distilled
  184. Non gaussianity
  185. Lens of indistinguishability
  186. Mm spatial
  187. Classifier free guidance
  188. Statistical inference
  189. Fast vision language models
  190. Guiding cross modal
  191. On inductive biases
  192. Tada
  193. Discrete neural flow
  194. Flexible language modeling
  195. Datasets documents repetitions
  196. Unigen enhanced training
  197. Instance optimality
  198. Follow the energy
  199. Apple foundation models tech report 2025
  200. Slowfast llava
  201. The super weight
  202. Investigating intersectional
  203. Pitch accent
  204. Uicoder
  205. Eliciting in context
  206. Optimizing contextual
  207. Interactive prototyping
  208. Optimal corpus
  209. Dicehubert
  210. Adaptive knowledge
  211. Accessibility project challenge
  212. Sequence to tree
  213. Ambisonics super resolution
  214. Affect models
  215. Conditioned video generation
  216. External validation
  217. Global calibration
  218. Mmau
  219. Mrakl
  220. Iclr 2025
  221. Aspera
  222. On the way
  223. Information geometry
  224. Symbolic regression
  225. New embedding spaces
  226. Pretraining data matches
  227. Vision transformers
  228. Overcoming vocabulary constraints
  229. Axlearn
  230. Iluvui instruction tuned
  231. Icml 2025
  232. Pts3d llm
  233. Quantspec
  234. Diffusion
  235. Self reflective
  236. Commutative vector quantization
  237. Naturalness
  238. Target concrete
  239. Addressing misspecification
  240. Apple intelligence foundation language models
  241. Scaling laws
  242. Geometries of truth
  243. Beyond sensor
  244. Private adversarial bandits
  245. Tracking the best
  246. Parameters flops scaling
  247. Learning to route
  248. Soup of experts
  249. Understanding input
  250. Super weight
  251. Distillation scaling laws
  252. Pruning large language
  253. Homomorphic encryption
  254. Cavia
  255. Evaluating long range
  256. Etva
  257. Advancing egocentric video
  258. Contrastive localized
  259. Towards safer ai agents
  260. Tic lm web scale
  261. Hypernetworks personalizing asr
  262. Synth4seg
  263. Private selection
  264. Revisiting uncertainty
  265. Phonetically augmented
  266. Normalizing flows
  267. Predicting preferences
  268. Flow matching
  269. Variational
  270. Trade offs
  271. Mobilebrick
  272. Pointersect
  273. Neural 3d relightable
  274. Collaborative machine learning
  275. Cvpr 2025
  276. Voice quality
  277. Chain of thought
  278. Proxy fda
  279. Beyond text compression
  280. Cross lingual transfer
  281. Prompting whisper
  282. World consistent
  283. Samplable anon aggregation
  284. Heart rate estimation
  285. Interleaved reasoning
  286. Voxelnet end to end
  287. Controlled synthesis
  288. Sync point drop
  289. Good stereoscopic
  290. Humanoid policy
  291. Cubify anything
  292. English accent
  293. Fastvlm efficient vision encoding
  294. Micronn on device
  295. Large photogrammetry model
  296. Sample complexity
  297. Local pan privacy
  298. Mechanisms of projective composition
  299. App store review
  300. How to verify distribution property
  301. Scaling diffusion language models
  302. Talking turns
  303. Dart denoising autoregressive transformer
  304. Tis dpo importance sampling
  305. Disentangled representational learning gromov
  306. Comotion concurrent 3d motion
  307. Diffusion elementary tutorial
  308. Scaling laws native multimodal models
  309. Ec dit
  310. No need to talk
  311. Focallens
  312. Length generalization transformers
  313. Exploring hallucinations
  314. 3d shape tokenization
  315. Modeling capabilities of language
  316. Dense to dynamic
  317. Task adaptive
  318. Ademamix optimizer
  319. Simple reflow
  320. Mm ego
  321. Relative contrastive learning
  322. Transporting activations
  323. Controlling language and diffusion
  324. Do llms know internally
  325. Adaptive batch size
  326. Ferret ui 2
  327. Large scale image caption
  328. Estimate uncertainty well
  329. Seedlm compressing
  330. Interpreting improving optimal control
  331. Mutual reinforcement llm dialogue
  332. Modeling speech emotion
  333. Universally instance optimal mechanisms
  334. Role of prosody
  335. Vibe visual analytics workflow
  336. Speaker ipl vector pseudo labels
  337. Ai sign language generation
  338. Easel promoting social emotional
  339. Interactive data augmentation
  340. Toolsandbox stateful conversational llm benchmark
  341. Evaluating text2sql solutions
  342. Univg diffusion model
  343. Exploring prediction targets
  344. Dr mpc deep residual
  345. Efficient streaming audio visual
  346. When does predictor know
  347. Kv prediction
  348. Multi rate residuals transformers
  349. Speech models using rank
  350. Selma speech enabled language
  351. Mm15 methods analysis insights
  352. Speech tokenization made simple
  353. Pixel space diffusion models
  354. Towards better instruction following
  355. Wearable accelerometer foundation models
  356. Grounding multimodal large
  357. Flex tok resampling
  358. Generalist embodied agents
  359. Efficient and effective
  360. Scalable graph neural network
  361. Armor egocentric
  362. Robust autonomy emerges
  363. Iwslt 2024 evaluation campaign
  364. Immerse diffusion generative
  365. Learning real world application
  366. Towards robust evaluation
  367. Set distribution networks
  368. Prediction preserving inputs
  369. Stable diffusion coreml apple silicon
  370. Sigmoid self attention
  371. Multimodal autoregressive
  372. Dataset decomposition
  373. Kaleido diffusion
  374. Cut your losses
  375. Adaptive training
  376. Mapping cells through time
  377. Transformation invariant
  378. Faster algorithms
  379. Step by step reasoning
  380. Spatially aware
  381. Reinforcement learning long horizon
  382. Follow instructions
  383. Instance optimal
  384. Heavy tails
  385. Estimate uncertainty
  386. Mitigating hallucinated translations
  387. Emotion expressive motion
  388. Compact neural tts
  389. Forget knowledge
  390. Higher order calibration
  391. Salsa soup
  392. Enhancing jepa
  393. Lazy algorithms
  394. Model based metrics
  395. Low bit
  396. Elegnt expressive functional movement
  397. Revisiting nonseparable binary
  398. Logic reasoning
  399. Delayed fusion integrating large
  400. Learning insulin glucose dynamics in the wild
  401. Dsplats 3d generation
  402. Robust hybrid learning
  403. Kg trick unifying
  404. Neural transducer training
  405. Interpreting clip insights
  406. Rephrasing the web
  407. Pre trained model
  408. Learning detect novel
  409. Motionprint
  410. Variable attention masking
  411. Transformers learn gradual
  412. Fingerprinting codes
  413. Rapid scalable
  414. Stochastic optimization
  415. Resource constrained
  416. Eelbert
  417. Autoregressive image models
  418. Veclip
  419. Parlor game arena
  420. Streaming anchor loss
  421. Merge vision
  422. Speakers speaking
  423. Corpus synthesis
  424. Large scale training
  425. Transport solvers
  426. Hybrid model learning
  427. Bin prediction
  428. Coml
  429. Investigating salient representation
  430. Leveraging large language models
  431. Acoustic model fusion
  432. Ctc based
  433. Supervised training
  434. Datacomp
  435. Smoothness induced
  436. Pretrained language models
  437. Hugs
  438. Fast optimal
  439. Adaptive weight
  440. Massively multimodal
  441. Fleek
  442. Deeppcr
  443. Tic clip
  444. Towards multimodal multitask
  445. Swap agnostic learning
  446. Planner
  447. Agnostically learning single
  448. Increasing coverage
  449. Flexible keyword
  450. User level differentially
  451. Vision inspired
  452. Relu
  453. Audio video learners
  454. Hyperdiffusion
  455. Slower respiration rate
  456. Fastvit
  457. Towards real world
  458. Automating behavioral testing
  459. Modality dropout
  460. Never ending learning
  461. Yield calibration
  462. Moonwalk
  463. Gender bias llm
  464. Gaudi
  465. Parameter free
  466. Nelf
  467. Reinforce data
  468. Parallel neural tts
  469. Conformalization sparse generalized
  470. Self supervised object goal
  471. Intelligent assistant language
  472. Sample size calculations
  473. Ferret
  474. Private online prediction
  475. Matching latent encoding
  476. Referring to screen
  477. Unified query rewriting
  478. Application agnostic language
  479. Latent phrase matching
  480. Monge map
  481. Resolving the mixing time
  482. Monge bregman occam
  483. Spatial librispeech
  484. Actionable data insights
  485. Nearest neighbor phrase
  486. Cross lingual knowledge transfer
  487. Roomdreamer
  488. Unified architecture
  489. Semi supervised long tailed
  490. Learning language specific layers
  491. Robustness multimodal learning
  492. Duet
  493. Open domain knowledgegraphs
  494. Self supervised temporal
  495. Near optimal algorithms
  496. From privacy robustness
  497. Machine translation
  498. Considerations distribution shift
  499. Entropy reconstruction
  500. Nerf
  501. Unconstrained channel pruning
  502. High throughput vector
  503. Multi stage diffusion
  504. Bootstrap own variance
  505. Autofocusformer
  506. Generalization on the unseen
  507. Helping machine translation
  508. Feedback effect
  509. Lip articulation
  510. Personalizing asr models
  511. Slick exploiting subsequences
  512. Mast
  513. Heimdal
  514. Improvements embedding
  515. User perceptions technical
  516. Ape
  517. Privacy computation trade offs
  518. Self supervision natural language
  519. Patch autoencoder
  520. Robust gan inversion
  521. Fastfill
  522. Efficient online augmentation
  523. Symbol guided hindsight
  524. Designing data
  525. Audio to intent
  526. Improving human annotation
  527. Diffusion probabilistic fields
  528. Loss minimization
  529. Unifying theory distance
  530. Shift curvature
  531. Beyond cage
  532. Joint speech transcription
  533. Active learning
  534. Segmentation guided
  535. Impact of language
  536. Homomorphic self supervised learning
  537. Rewards encoding
  538. Non autoagressive neural machine
  539. Treatise on fst lattice
  540. Prompting conversation
  541. Soft pseudo labeling
  542. Large scale observational
  543. Elastic weight consolidation
  544. Analyzing mitigating repetitions
  545. Subspace recovery
  546. Modeling spoken information queries
  547. Learning bias reduced
  548. Modeling heart rate
  549. Statistical deconvolution
  550. Privacy noisy stochastic
  551. Isotropic networks
  552. Calibration generalization gap
  553. Low rank optimal transport
  554. Generalization unseen data boolean
  555. Multi view bootstrapping
  556. Latent temporal flows
  557. Flair
  558. Photoplethysmography motion sensor fusion
  559. Safe real world
  560. Improving generalization
  561. Detector free image
  562. Open response surveys
  563. Physiomtl
  564. Toward supporting quality
  565. Mean estimation
  566. Two layer bandit
  567. Neural incident light field
  568. Generative multiplane images
  569. Combining compressions
  570. Improving voice trigger
  571. Multi objective hyper parameter
  572. High performance library
  573. Taxonomy overfitting
  574. Device directed speech
  575. Vision based hand
  576. Minimax demographic group
  577. Generating textures
  578. Private frequency estimation
  579. Learning to reweight data
  580. Regularized training
  581. Reinforcement learning robotics
  582. Optimal algorithms
  583. Integrating categorical features
  584. Position prediction
  585. Reachability embeddings
  586. Space efficient representation
  587. Self conditioning pre trained
  588. Training a tokenizer
  589. Investigating model representations
  590. Leveraging entity representations
  591. Efficient representation
  592. Dynamic memory
  593. Forward compatible training
  594. Unsupervised learning
  595. Enabling hand gesture
  596. Low resource adaptation
  597. Icon labelling
  598. Cinematic l1 video stabilization
  599. Bytes are all you need
  600. Streaming on device
  601. Vision inspired method
  602. Differentiable k means
  603. Vocal effort modeling
  604. Learning compressed embedding
  605. Private vector mean estimation
  606. Time sensitive finetuning
  607. Continuous construction
  608. Llm fusion low rank
  609. Evaluating iwslt2023 tasks
  610. Server side rescoring
  611. Espnet spk
  612. Agrame
  613. Embedding pose graph
  614. Vision transformer
  615. Learning based 3d reconstruction
  616. Neural fisher
  617. Screen relationships
  618. Efficient diffusion models
  619. Odgen
  620. Matryoshka diffusion models
  621. Facial motion synthesis
  622. Neural face video
  623. Swallowing the bitter pill
  624. Automatic creative selection
  625. Synthetic defect generation
  626. Composing interactive interfaces
  627. Gain propagation
  628. Mgie
  629. Generative modeling
  630. Poly view contrastive learning
  631. Generalizing confusion matrix
  632. End to end
  633. Differential secrecy
  634. Dynamic view synthesis
  635. Boot
  636. Contrastive recurrent state space models
  637. Learning spatiotemporal occupancy
  638. Fair sa
  639. Federated personalization
  640. Consistent collaborative filtering
  641. Hindsight priors reward learning
  642. Open domain question answering
  643. Arkitscenes
  644. Batchquant
  645. Rim
  646. Mmiu visual intent
  647. Self supervised semi supervised learning
  648. Self supervised supervised methods
  649. Stochastic contrastive learning
  650. Reconstructing training data
  651. High fidelity 3d reconstructions
  652. Finding experts
  653. Breiman two cultures
  654. Pause information
  655. Entity knowledge conflicts
  656. On device neural speech
  657. Enforcing fairness
  658. Learning compressible subspaces
  659. Cross domain data integration
  660. Deep partial point cloud registration
  661. Multitask learning with cross attention
  662. Lidar segmentation
  663. Screen parsing
  664. User initiated repetition based recovery
  665. Audiovisual synthesis using tacotron2
  666. Managing ml pipelines
  667. Survey on privacy
  668. Deep encoding
  669. Generation of synthetic geospatial images
  670. Bilingual end to end asr
  671. Transformers learn
  672. Online automatic speech recognition
  673. Nonparametric differentially private median confidence intervals
  674. Memorization of irrelevant training
  675. Retrievalfuse
  676. Neural networks for intrinsic image decomposition relighting
  677. Subject aware contrastive learning
  678. Private adaptive gradient methods
  679. Lossless compression
  680. Implicit greedy rank
  681. Streaming transformer
  682. Bootleg self supervision
  683. Analysis and tuning
  684. Smooth sequential optimization
  685. Interdependent variables
  686. Leveraging query resolution
  687. Spatio temporal context
  688. Morphgan
  689. Private stochastic convex optimizations
  690. Probabilistic attention for interactive segmentation
  691. Attention free transformer
  692. Learning to optimize black box evaluation metrics
  693. Instance level task parameters
  694. Evaluating entity disambiguation amber
  695. Combined resolution of ellipses
  696. Visual cues in audiovisual speech enhancement
  697. Extracurricular learning
  698. Learning sample adaptive policy
  699. Noise robust named entity
  700. Dynamic curriculum learning
  701. Multimodal punctuation prediction
  702. Discriminative entity aware language
  703. Progressive voice trigger
  704. On device false trigger
  705. Training dnn hmm
  706. Error driven pruning
  707. Unconstrained scene generation
  708. Joint speech recognition and translation
  709. Recommending accessibility
  710. Video frame interpolation
  711. Neural feature selection
  712. Frame level specaugment
  713. Neural tts long form
  714. Creating accessibility metadata
  715. Question rewriting
  716. Lombard neural speech synthesis
  717. Stuttering event detection
  718. Generating natural questions
  719. Enhanced direct delta mush
  720. Offline reinforcement learning
  721. Hypersim
  722. Individual privacy accounting
  723. What neural networks memorize
  724. Learning neural network subspaces
  725. Privacy amplification by shuffling
  726. Wearable ecg recordings
  727. Generating synthetic images
  728. Improving human labeled data
  729. Tract
  730. Layer wise data free cnn compression
  731. Stochastic gradient descent
  732. Multichannel voice trigger
  733. Collegial ensembles
  734. Utilizing imperfect synthetic
  735. Downbeat tracking with tempo
  736. Learning soft labels
  737. Axnav
  738. Slingshot mechanism
  739. Critical regularizations
  740. Tensor programs
  741. Finerecon
  742. Robust joint shape
  743. Dnikit
  744. Fine tuning strategies
  745. Faster nerf
  746. Model stability with continuous data
  747. Multi view stereo
  748. Ddim
  749. Efficient multimodal neural networks
  750. Challenges adversarial image
  751. Randomized controlled trials
  752. Neural human radiance field
  753. Masked auto encoder
  754. Naturalistic head motion
  755. Comparison of question rewriting methods
  756. Rescribe authoring and automatically editing audio descriptions
  757. Open domain question answering goes conversational
  758. Neural machine translation
  759. Making smartphone augmented reality apps accessible
  760. Automatic class discovery one shot interactions acoustic activity
  761. Question reformulation and answer selection
  762. Task agnostic data augmentation
  763. Mage fluid moves between code
  764. Data parameters
  765. Capsules with inverted dot product attention routing
  766. Jointly learning to align and translate with transformer models
  767. Speech translation
  768. Improving knowledge base construction from robust infobox extraction
  769. Adversarial fisher vectors for unsupervised representation learning
  770. Exploration of data augmentation and sampling techniques
  771. Voice trigger detection from lvcsr hypothesis lattices using bidirectional lattice recurrent neural networks
  772. Parametric cepstral mean normalization for robust speech recognition
  773. App usage predicts cognitive ability
  774. Lyric document embeddings
  775. Learning invariant representations
  776. Cease
  777. Data resource efficient
  778. Diffusion fields
  779. Randomized algorithms
  780. Range regularization
  781. Controllable music
  782. Query auto completion email
  783. Semand
  784. Delphi
  785. Steer
  786. Marrs
  787. Compress compare
  788. Generating gender alternatives
  789. Private personalized histogram
  790. Asr contextualization
  791. Mobileone
  792. Instruction following speech
  793. Pointconvformer
  794. Subspace disentanglement
  795. Kglens towards efficient
  796. Synthetic query gen llm
  797. Improved modelling federated datasets
  798. Conformer based speech
  799. Data incubation
  800. Hdr environment map estimation
  801. Generalized reinforcement meta learning few shot optimization
  802. Affine based attention
  803. Clip model zoo experts
  804. Mkqa
  805. Conversational semantic parsing
  806. Recipe for compute
  807. Comparative analysis personalized voice
  808. Wally search
  809. Least squares binary quantization neural networks
  810. Pro tip
  811. Spinning configurable scalable
  812. Equivariant neural rendering
  813. Knowledge transfer
  814. Armada augmented reality
  815. Biscuit scaffolding llm
  816. Contextq
  817. Pfl research
  818. Complex spatial datasets
  819. Conformal prediction
  820. Variance adaptive thompson sampling
  821. Kv runahead
  822. Compressing llms
  823. Superposition prompting
  824. Datacomp lm search
  825. Mofi
  826. Model driven heart
  827. Jointnet
  828. Transformers reason abstract symbols
  829. Label efficient sleep staging
  830. Direct 2 5
  831. Llms as generalizable policies
  832. Dynamic token pruning
  833. Intelligent visual deductive reasoning
  834. Fixally accessibility
  835. Talaria
  836. Openelm
  837. Frequency aware masked autoencoders
  838. Slingshot effect
  839. Vanishing gradients reinforcement
  840. Kpconvx
  841. Unified bayesian
  842. Fedhyper
  843. Privacy preserving quantile
  844. Overcoming pitfalls vision language
  845. Dense material
  846. Population expansion
  847. Hierarchical dynamic prompt
  848. Combining speakers multiple language
  849. Recurrent drafter
  850. Retrieval asr
  851. Direct large language
  852. Fast and explicit neural view synthesis
  853. Fed learning diff privacy
  854. Gender bias transfer
  855. Mia bench
  856. Ui intent
  857. Towards automated accessibility report
  858. Conditional loradapter
  859. Positional description
  860. How far can transformers
  861. Private density estimation
  862. Optimizing byte representation for asr
  863. Optimization without retraction
  864. Gsm symbolic
  865. Error resistance
  866. Novel view
  867. Data centric rlhf
  868. Asr n best rescoring
  869. Differentially private samplers
  870. Improving gflownets
  871. Pixel hierarchies
  872. Careful with that scalpel
  873. Realm reference
  874. Repcnn micro
  875. Remove downstream
  876. Acl pseudo labeling
  877. How smooth is attention
  878. Ferretv2
  879. Neural implementation breniers
  880. Entity disambiguation fusion decoding
  881. Whispering experts
  882. Posterior uncertainty quantification
  883. Ape active prompt engineering
  884. Codeact
  885. Minimal degree bias
  886. Computationally efficient calibration
  887. Sam clip
  888. Contrasting multiple representations
  889. Multi gyroscope calibration
  890. Modality dropout for improved performance
  891. Non verbal sound
  892. Hierarchical prosody modeling
  893. Hybrid transformer and ctc networks
  894. Improving on device speaker
  895. Stacked 1d convolutional networks
  896. Momentum approximation
  897. Omnipredictors for regression
  898. Class lm and word mapping
  899. Controllable neural text to speech synthesis
  900. Complementary language model and parallel bi lrnn
  901. Learning to rank intents in voice assistants
  902. Structured control nets for deep reinforcement learning
  903. Functional narratives
  904. Double talk robust multichannel accoustic echo
  905. Robust multichannel linear prediction
  906. Improving discrete latent representations with differentiable approximation bridges
  907. Learning to branch multi task learning
  908. Projected language models
  909. Collaborative agents
  910. Convkgyarn datasets
  911. A generative model joint natural language understanding generation
  912. Lower bounds for locally private estimation via communication complexity
  913. Single training dimension selection for word embedding with pca
  914. Empirical evaluation of active learning techniques for neural mt
  915. Toad
  916. Mirroring to build trust in digital assistants
  917. Multiple futures prediction
  918. Reverse transfer learning can word embeddings trained for different nlp tasks improve neural language models
  919. Developing measures of cognitive impairment in the real world from consumer grade multimodal sensor streams
  920. Model compatibility
  921. Raise to speak an accurate low power detector for activating voice assistants on smartwatches
  922. Towards learning multi agent negotiations via self play
  923. Addressing the loss metric mismatch with adaptive loss alignment
  924. Leveraging user engagement signals for entity labeling in virtual assistant
  925. Foundationdb record layer a multi tenant structured datastore
  926. Covid mobility
  927. Pine
  928. Nonlinear conjugate gradients for scaling synchronous distributed dnn training
  929. Acoustic neighbor embeddings
  930. Element level differential privacy
  931. Coarse to fine optimization for speech enhancement
  932. Data to text
  933. Implicit acceleration
  934. Context tuning retrieval
  935. Efficient 3dim
  936. Variational saccading efficient inference for large resolution images
  937. How easy
  938. New benchmark and progress toward improved weakly supervised learning
  939. Private and personalized
  940. Protection against reconstruction and its applications in private federated learning
  941. Stochastic weight averaging in parallel
  942. Connecting and comparing language model interpolation techniques
  943. Computational bottlenecks
  944. Understanding visualizing data iteration machine learning
  945. Stabilizing transformer training
  946. Filter distillation for network compression
  947. Learning from simulated and unsupervised images through adversarial training
  948. Progressive transport
  949. Tempura
  950. Multi task learning speaker verification
  951. Sndcnn
  952. Generating multilingual voices
  953. Accurate knowledge distillation
  954. Efficient large language
  955. Multi task learning voice trigger detection
  956. Improving language identification multilingual speakers
  957. Scalable multilingual frontend tts
  958. Variational neural machine translation with normalizing flows
  959. Handwritten chinese character recognition
  960. Vision model
  961. Contrastive alignment instructions
  962. Smart audit
  963. Relu strikes back
  964. Divide conquer
  965. Quality estimation data annotation
  966. Towards time
  967. Visual recognition accuracy
  968. Mobileclip
  969. Mel spectrogram
  970. Device directed
  971. Implicit bias
  972. Llm device directed speech detection
  973. Memory retaining
  974. Detecting emotion primitives
  975. Tic clip v2
  976. Ferretui mobile
  977. Construction paired knowledge graph
  978. Duo llm
  979. Human motion understanding
  980. Predictor corrector
  981. World english language model
  982. Leveraging acoustic cues
  983. Leveraging periodicity
  984. Model compression in practice
  985. Contextual bandits
  986. Adascale sgd
  987. Worst cases policy gradients
  988. Skip clip self supervised spatiotemporal representation learning by future clip order ranking
  989. Wasserstein flow
  990. Exploring retraining free speech recognition for intra sentential code switching
  991. Monge displacements
  992. Learning conditional error model for simulated time series data
  993. Error modeling
  994. On modeling asr word confidence
  995. Speaker independent speech driven visual speech synthesis using domain adapted acoustic models
  996. Overton
  997. Unsupervised style content separation
  998. Modeling patterns of smartphone usage and their relationship to cognitive health
  999. Transfer learning pruning
  1000. Label shift estimators
  1001. Data platform machine learning
  1002. Modal representations
  1003. Lattice based improvements for voice triggering
  1004. Bandwidth embeddings
  1005. Scaling smart
  1006. Clip experts
  1007. Revealing utilized rank
  1008. Gesture customization
  1009. Active learning for domain classification
  1010. Generating molecular
  1011. Sensing linear probing
  1012. Predicting entity popularity to improve spoken entity recognition by virtual assistants
  1013. Sliced wasserstein discrepancy for unsupervised domain adaptation
  1014. Enhancing ctc speech
  1015. Scale em
  1016. Write hypothesis
  1017. Reward generalization
  1018. Cultural translation
  1019. Multicalibration necessity
  1020. One wide ffn
  1021. Single stage diffusion
  1022. Aggregate and adapt
  1023. Item to item recommendation
  1024. Neural network based modeling of phonetic durations
  1025. Consistent collaborative filtering updates
  1026. Continuous pseudo labeling
  1027. Livepose
  1028. Large vocabulary
  1029. Differentially private heavy
  1030. Human following
  1031. Efficient convbn
  1032. Synthdst
  1033. Mm1 methods analysis insights
  1034. Applying rlaif
  1035. Data filtering networks
  1036. Leveraging gans continuous path keyboard
  1037. Speech is more
  1038. Redrafter nvidia tensorrt llm
  1039. Personalized heartrate
  1040. Core ml on device llama
  1041. Gender alternatives
  1042. Mobile applications accessible
  1043. Recognizing people photos
  1044. Ml compute training on mac
  1045. Inverse text normal
  1046. Panoptic segmentation
  1047. Siri voices
  1048. Bridging the domain gap for neural models
  1049. Personal voice
  1050. Scenes differential privacy
  1051. Speech recognition
  1052. Learning to generate radiance fields
  1053. On device scene analysis
  1054. Latent language diffusion model
  1055. Humanizing wer
  1056. Federated learning speech
  1057. Gan
  1058. Voice trigger
  1059. Face detection
  1060. Neural engine transformers
  1061. Salient object segmentation
  1062. Roomplan
  1063. Improving neural subspaces
  1064. Modeling covid mobility
  1065. Interpretable adaptive optimization
  1066. Rgb x classification
  1067. Handwriting
  1068. Personalized hey siri
  1069. Optimizing siri on homepod in far field settings
  1070. Learning with privacy at scale
  1071. Language identification from very short strings
  1072. Cross initialization
  1073. Hey siri
  1074. Regionally specific language models
  1075. Can global semantic context improve neural language models