AI
Beyond Suspicious Steps: Ontological Trust in Long-Horizon Agents
Researchers have proposed a new method for monitoring the behavior of long-horizon agents, which are AI systems that operate across multiple steps and...
AI
AI
Evaluating the Diversity of AI-Generated Content with Diversity Profiles
Researchers have proposed a new way to evaluate the diversity of AI-generated content called 'diversity profiles.' These profiles are curve-valued sum...
AI
Neuro-symbolic learning over OWL 2 DL via consequence-based compilation to differentiable circuits
Researchers have developed a new method for combining symbolic and neural network-based learning. They call it Baobab, which translates OWL 2 DL ontol...
AI
The Curious Case of Exploding DecPOMDPs: Containing the Fire through Policy Counting
Researchers have developed a new approach to tackle the complexity of multi-agent decision making under uncertainty. They propose counting policies in...
AI
D$^2$ACCI: A Dual-Loop Diagnostic Protocol for Evidence-Preserving Agent Memory
Researchers have proposed a new protocol called D$^2$ACCI to improve the diagnosis of failures in large language model (LLM) agents' memory systems. T...
AI
StartupBench: Benchmarking General-Purpose Agents on Market-Validated End-to-End Workflows
Researchers have introduced StartupBench, a benchmark for general-purpose agents that evaluates their ability to complete real-world tasks. Unlike exi...
AI
ARASH: Adaptive Retrieval And Shot Selection for Tabular Prediction
Researchers have developed ARASH, a method to improve the efficiency of Tabular Foundation Models (TFMs) in tabular prediction tasks. These models are...
AI
AutoResearch: Insight In, Hallucination Out
Researchers have developed AutoResearch, a two-stage system that aims to ensure the reliability of research processes. The system combines emerging re...
AI
Adaptive Policy Portfolios for Robust Markov Decision Processes
Researchers have developed a new approach to creating 'adaptive policy portfolios' for Markov decision processes. These portfolios are sets of pre-com...
AI
EvoTS-Agent: A Self-Evolving LLM Agent for Financial Time Series Change Point Detection
Researchers have developed EvoTS-Agent, a self-evolving language model agent that can detect changes in financial time series data without human inter...
AI
Procedural Content Metageneration via Program Search and Continual Abstraction Discovery
Researchers have developed a method to generate procedural content, such as levels in video games, using large language models and evolutionary progra...
AI
Towards Zero-Shot Task Transfer with Neurosymbolic World Models
Researchers propose a new type of world model that combines neural networks with symbolic components. This allows for zero-shot task transfer, meaning...