Home
Technology
Gaming
Entertainment
World & Business
Science
Sports
AI

World & Business

Home
Technology
Gaming
Entertainment
World & Business
Science
Sports
AI

WEDGE proposes training worlds shaped by navigation agents' failures · Digg

Announcement

WEDGE proposes training worlds shaped by navigation agents' failures

WEDGE's creators say it identifies where each agent struggles and generates learnable environments targeting those weaknesses.

TLDR

The team behind WEDGE says navigation agents can fail under different environmental conditions, even on the same task. Its framework uses each agent's failures to generate learnable training environments aimed at its weaknesses, then reassesses the agent after each round of adaptation to guide the next round.

Combined views

405

6 Sources, first seen 2h ago

Combined views

405

6 Sources, first seen 2h ago

15 likes

2h ago

first seen 2h ago

15 likes

1 saves

16 reposts

1 saves

16 reposts

Featured Source

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

Sentiment

Positive——Negative

Summary

Not enough discussion yet.

No sentiment analysis available yet.

6 Sources

@zhan1624🚨 Excited to share WEDGE: From Failures to Learnable Worlds, our new work on policy-conditioned environment generation for navigation agents! We found that even on the same navigation task, agents tend to fail under different environmental conditions. Yet most environment generators do not consider what each agent actually needs to improve. 🤔 This led us to a question: Can we use an agent's own failures to guide the generation of training environments? ➡️ We introduce WEDGE, a framework where an agent's failures shape the worlds it trains in next. WEDGE identifies the conditions under which each agent struggles and generates new environments that target those weaknesses while remaining learnable. After each round of adaptation, it re-diagnoses the agent to guide the next round of environment generation. Thread🧵👇2h

@EliasEskinRT @zhan1624: 🚨 Excited to share WEDGE: From Failures to Learnable Worlds, our new work on policy-conditioned environment generation for na…2h

@ZunWang919The value of a training environment depends on what the agent still needs to learn. In WEDGE, we use navigation agents’ failures to generate worlds that target their weaknesses while remaining challenging but learnable.2h

@mohitban47RT @ZunWang919: The value of a training environment depends on what the agent still needs to learn. In WEDGE, we use navigation agents’ fai…1h

@codezakhIn our newest work on environment generation, we discover environment features (as executable functions) that explain why an embodied navigation agent fails, and use those features to generate novel learnable environments where the policy gets practice with the conditions that make it fail. Over multiple rounds, we can discover new weaknesses and keep generating new environments! 🧵👇46m

6 Sources

@zhan1624🚨 Excited to share WEDGE: From Failures to Learnable Worlds, our new work on policy-conditioned environment generation for navigation agents! We found that even on the same navigation task, agents tend to fail under different environmental conditions. Yet most environment generators do not consider what each agent actually needs to improve. 🤔 This led us to a question: Can we use an agent's own failures to guide the generation of training environments? ➡️ We introduce WEDGE, a framework where an agent's failures shape the worlds it trains in next. WEDGE identifies the conditions under which each agent struggles and generates new environments that target those weaknesses while remaining learnable. After each round of adaptation, it re-diagnoses the agent to guide the next round of environment generation. Thread🧵👇2h

@EliasEskinRT @zhan1624: 🚨 Excited to share WEDGE: From Failures to Learnable Worlds, our new work on policy-conditioned environment generation for na…2h

@ZunWang919The value of a training environment depends on what the agent still needs to learn. In WEDGE, we use navigation agents’ failures to generate worlds that target their weaknesses while remaining challenging but learnable.2h

@mohitban47RT @ZunWang919: The value of a training environment depends on what the agent still needs to learn. In WEDGE, we use navigation agents’ fai…1h

@codezakhIn our newest work on environment generation, we discover environment features (as executable functions) that explain why an embodied navigation agent fails, and use those features to generate novel learnable environments where the policy gets practice with the conditions that make it fail. Over multiple rounds, we can discover new weaknesses and keep generating new environments! 🧵👇46m

Today's Rank

—

Not ranked yet

Today's Rank

—

Not ranked yet