已审阅并标注「模仿学习」的研究与资产。
条目依据已审阅标签归类，同一主题不表示资源之间存在依赖关系。日期保留来源精度，注释保留原语言。
WildAWM synthesizes resettable GUI/MCP/CLI environments and cross-interface tasks, using execution-based checks of solvability, difficulty and verifier consistency before supervised and reinforcement learning.
ScopeMem conditions latent-memory prefixes on a frozen GUI agent's initial action and trains them to correct errors while preserving correct decisions. Its evaluation separates action repairs from harmful revisions.
FocusGUI learns step-specific weights over screenshot, audio, video and interaction-context channels, injecting a compact residual into a frozen decoder without pruning its inputs.
CUA-Lite unifies environment interfaces, supervised data formats and per-model harnesses for evaluation, reinforcement learning and supervised fine-tuning across desktop, browser and mobile tasks.
Separates next-screen prediction from action evaluation in GUI lookahead and trains Direct Action Ranking on reference actions, comparing prediction-augmented scoring under matched offline supervision.
WebMix uses a scout-produced interface manual to route mixtures of source-trained LoRA adapters for web agents, adapting to target interfaces without target-time gradient updates.
ECHOES learns recursive textual task memory through privileged screenshot-history distillation and GRPO, then predicts GUI actions from the current screenshot and memory.
GUI-Hopper combines verified Android deeplinks with GUI actions, using landing-screen descriptions, hybrid trajectory training and reinforcement learning to select navigation shortcuts and recover through GUI control.
WebMRE pairs guidance sentences with grounded actions in an offline benchmark derived from successful WebArena traces. Controlled decoding-prefix interventions test how guide content affects action selection, while separating offline prediction metrics from live browser task success.
GUI-SD-v2 extends on-policy self-distillation from grounding to multi-turn GUI interaction. It first strengthens responses to privileged guidance, then selectively distills step-specific reasoning and memory support for later actions.
MaP unifies GUI-navigation objectives by masking and predicting components of multi-turn trajectories. Role-aware adapters address heterogeneous training data, and evaluations compare the approach with direct task-mixture training across navigation benchmarks.
LLaDA-UI adapts block-wise diffusion generation to a vision-language GUI agent through multimodal pretraining and GUI supervised fine-tuning. It evaluates grounding and navigation across multiple platform settings.
ASIL exposes application state as structured JSON and provides code-executable semantic actions. Its application benchmark compares this interface with screenshot control and application-native APIs, and studies supervised and reinforcement learning on the structured interface.
LookAgain treats a GUI coordinate as a revisable hypothesis, rendering its marker and a local screenshot patch before confirmation. Reflective supervised trajectories provide a cold start, followed by GRPO using terminal grounding correctness.
Gated Hindsight Distillation gives a teacher access to the next screenshot during training while the student sees only the available trajectory prefix. Distillation is applied when hindsight repairs a student failure, and the resulting policies are evaluated on Android tasks.
AppDeltaWorld predicts executable code updates for the next mobile UI state, using transition-constrained retrieval and rendered visual assets. It evaluates reconstruction fidelity separately from policy training and adaptation using generated environments.
StepReflect predicts structured per-action reflections from transition specifications and paired visual evidence. The paper distinguishes offline transition-label accuracy from online task success when the reflection model is integrated into mobile agents.
Qwen-CUA operates through screenshots and keyboard/mouse events while compressing older visual history into reusable context blocks. Its report combines verifiable task generation, trajectory-based training and separate browser and Bash-augmented studies.
MAGA distills specialized GUI agents into one policy by allocating supervision according to structured-action correctness rather than treating all response tokens equally. A training-only teacher hint supports fusion across mobile, web and desktop settings without changing student inputs.
Introduces Uni-GUI, a collection of executable desktop and mobile trajectories, and UI-MOPD, which trains a shared policy from its own rollouts using platform-specific teachers. The distillation procedure routes each visited state to the appropriate teacher to preserve distinct interaction conventions.
PEEU explores web environments and turns hindsight experience into aligned high-level planning supervision. A hierarchical task-decomposition analysis examines why atomic-skill competence alone does not ensure planning generalization across websites.
WinDOM obtains bounding-box supervision from a Windows 11 web reimplementation driven through Playwright, and studies rejection-sampling distillation for small GUI grounders. It treats the depth of the supervised cold start as a GRPO choice, comparing same-size EMA and larger same-family teachers.
MemGUI-Agent makes context management an action chosen by the same policy that controls the mobile UI. Structured history, UI-state and recent-step fields are trained from annotated trajectories to retain cross-app facts while limiting context growth.
Skill-Guided Continuation Distillation first lets a policy reach its own off-trajectory states, then uses skill guidance to collect successful continuations. Mixing those continuations with expert traces supplies supervision for recovery states absent from demonstrations.
Proposes quality-aware self-distillation for GUI coordinate generation under student-produced prefixes. A correctness-aware gate suppresses coordinate-token supervision that cannot reach the target box, while teacher-probability scaling adjusts the remaining signal strength.
ProCUA-SFT generates tasks and synthetic trajectories on live desktops seeded with document content, checking task preconditions before rollout. The resulting step-prefix examples are formatted to match inference context and used for supervised GUI-agent fine-tuning.
Speculative Rollback Correction lets a student web agent execute short branches before teacher review localizes harmful deviations. Verified continuations and localized corrections feed supervised training while an archive retains diverse successful solutions.
AliyunConsoleAgent trains browser agents to verify documented procedures against changing cloud-console UIs through trajectory distillation and online GRPO. Provisioned rollout environments and backend audit-log checks support training and outcome evaluation.
Weasel selects web-agent training steps by balancing importance and diversity across states, sites and interaction patterns. It also prunes training accessibility trees around known action targets and regenerates rationales to match the model’s reasoning style.
ToolCUA synthesizes interleaved GUI/tool trajectories and trains agents to choose between graphical actions and higher-level tools. Staged supervised and reinforcement learning culminate in online hybrid-environment training with a reward for efficient execution paths.
Introduces Mem-W, which compresses previous trajectories and current-session segments into latent memory tokens used directly by the GUI policy. Self-distillation and outcome supervision train this memory representation for web and mobile navigation.
AutoSurfer explores website actions breadth-first, retaining page and trajectory knowledge to guide task synthesis and trajectory refinement. It evaluates the resulting training data by fine-tuning a web agent on WebArena and examining synthesized-task diversity.
Presents OpenMobile, which builds exploration-based environment memory to synthesize task instructions and switches between learner and expert policies during trajectory collection. The resulting data includes recovery behavior and is tested through mobile-agent training and benchmark-overlap analysis.
WebAgentGuard separates prompt-injection detection from the operating web agent through a parallel multimodal guard. The guard is trained on synthetic visual/textual examples using reasoning-focused supervised fine-tuning followed by reinforcement learning.
EE-MCP studies application-dependent mixtures of GUI actions and MCP calls using generated environments, trajectory collection and filtered training. An experience bank extracts rules from trajectory comparisons for inference-time improvement, distinct from model distillation.
Presents MolmoWebMix, combining synthetic browser trajectories, human demonstrations and GUI perception data, and MolmoWeb screenshot-to-action policies trained from it. Browser benchmarks also test parallel rollouts with best-of-N selection.
Introduces a two-stage mobile GUI learner that first co-evolves trajectories and policies through rejection fine-tuning. Group Relative Self-Distillation then uses successful rollout branches to provide step-level supervision for failed trajectories, evaluated on AndroidWorld.
Introduces a pipeline that filters individual steps from noisy computer-agent rollouts and augments them with reasoning for training. WebSTAR and WebSCORE support GUI-policy supervision and a distilled process reward model, StepRM.
Introduces SynthAgent, which generates tasks by exploring website elements and refines them when observations reveal conflicts. A later trajectory-refinement stage removes misaligned behavior before fine-tuning agents for the target websites.
Trains BrowserAgent to use Playwright-based scrolling, clicking and typing directly on webpages for information-seeking tasks. Supervised and rejection fine-tuning are paired with explicit memory for conclusions carried across steps.
AutoPlay explores application states and functions to generate tasks, then uses a model executor and verifier to construct demonstrations. It produces 20,000 tasks across 20 Android apps and 10,000 tasks across 13 Ubuntu apps for supervised and reinforcement learning of UI agents.
RAG-GUI trains a small guidance adapter to filter and summarize retrieved web tutorials while keeping the acting agent fixed. Supervised and rejection-sampling stages prepare the adapter for inference-time assistance, tested on offline web/mobile tasks and interactive Android tasks.
UIPro pretrains GUI understanding on 20.6 million tasks and then harmonizes interaction datasets through a unified action space. Continued fine-tuning uses the merged multi-platform data to train action prediction for generalist GUI operation.
Introduces OmniActor, which shares shallow model layers while separating deeper parameters to train jointly on GUI and embodied interactions. A unified action representation and mixed training data support evaluation in both task families.
CODA combines a generalist planner with a specialist GUI executor and trains application-specific planners through decoupled GRPO. Successful specialist trajectories then support supervised training of a consolidated planner, evaluated across four ScienceBoard applications.
SEAgent explores unfamiliar software using a world-state assessor and an automatically generated task curriculum. It learns from failed and successful interactions through adversarial imitation and GRPO, then combines specialist experience into a generalist agent evaluated on five OSWorld software environments.
Studies compute allocation for web-agent post-training with teacher-generated supervised fine-tuning followed by on-policy reinforcement learning. It samples 1,370 training configurations and uses bootstrap analysis to examine hyperparameter sensitivity on WorkArena and MiniWoB++.
AppVLM first learns app actions from AndroidControl and then refines its policy through data collected in AndroidWorld. The study compares offline action prediction and online task completion for a lightweight vision-language smartphone agent.
Synatra converts indirect procedural knowledge, including human-oriented tutorials, into direct demonstrations for digital agents. It trains a 7B CodeLlama on 100,000 synthesized demonstrations and evaluates the resulting policy on Mind2Web, MiniWoB++, and WebArena.
Introduces GLAINTEL, a Flan-T5-based product-search web agent studied under unsupervised learning, demonstration-based training, and unsupervised domain adaptation. It compares ways to combine demonstrations with reinforcement learning.
Trains WebGUM offline from demonstrations by jointly tuning an instruction-following language model and a vision encoder over webpage screenshots and HTML. The agent predicts browser actions and is evaluated on MiniWoB, WebShop, and Mind2Web.
Introduces PUMICE, an agent that learns mobile-app procedures and conditional concepts through natural-language conversation and GUI demonstrations, recursively resolving ambiguous instructions with users.
A federated mobile-agent training and evaluation suite that varies user-data heterogeneity and application distribution, with released partition annotations and inference/scoring scripts.
Published AgentSynth task definitions and recorded GUI trajectories, with OSWorld and InSTA training configurations and packaged screenshots, thoughts and commands.
Conversation-formatted supervision released for fine-tuning AutoDroid V2, with the project documenting a ShareGPT-style JSON configuration.
Flattened web-interaction training records with accessibility-tree observations, assistant completions, trajectory rewards and step-link metadata.
A published JSONL corpus of synthesized planning and causal knowledge for GUI-agent mid-training, linked directly from the official GUI-CIDER project.
A TongUI-collected training corpus from multimodal web tutorials, distributed as annotation files and split screenshot archives for GUI trajectory supervision.
A released GUI-agent mid-training mixture and supporting domain datasets, together with a post-training trajectory file and image archives for web and mobile interaction.
Video-derived mobile navigation annotations that pair scene boundaries and screen or element boxes with action labels for training and evaluating GUI agents.
第 1–60 条，共 84 条资源。
通过主题导出清单获取完整记录和证据。克隆目录后，可将主题记录直接交给标准 RAG 与引用导出工具。
