Introduces least-to-most prompting, combining problem decomposition and step-by-step solving, greatly enhancing large language models' reasoning on complex tasks.

Denny Zhou, Nathanael Schärli, Le Hou et al.

Gato, a 1.2B-parameter Transformer, unifies multi-modal, multi-task learning, achieving superior performance across 604 diverse tasks including Atari, robotics, and language.

Scott Reed, Konrad Zolna, Emilio Parisotto et al.

ESCM$^2$ model uses counterfactual risk minimization to address CVR estimation bias, significantly enhancing recommender system performance.

Hao Wang, Tai-Wei Chang, Tianqiao Liu et al.

Proposes adversarial motion priors (AMP) using limited motion capture data to replace complex reward functions for natural, energy-efficient robot locomotion.

Alejandro Escontrela, Xue Bin Peng, Wenhao Yu et al.

Formalizes causal sufficient, counterfactual, and actual causation definitions to enhance action-guiding XAI, improving fairness and interpretability.

Sander Beckers

Proposes goal-conditioned RL algorithms with goal relabeling, reward shaping, and model-based planning, achieving 20-30% efficiency gains.

Minghuan Liu, Menghui Zhu, Weinan Zhang

Proposes Co-12 properties for systematic evaluation of 300+ XAI papers, emphasizing quantitative metrics.

Meike Nauta, Jan Trienes, Shreyasi Pathak et al.

Proposes MISLID, an adaptive fixed-confidence Top-m algorithm for misspecified linear bandits, matching the theoretical lower bound.

Clémence Réda, Andrea Tirinzoni, Rémy Degenne

ESAM introduces stochastic weight perturbation and sharpness-sensitive data selection, reducing computational cost by 60% while maintaining or improving generalization performance.

Jiawei Du, Hanshu Yan, Jiashi Feng et al.

ASDiv dataset enhances math word problem solver evaluation accuracy through diverse lexicon and problem types.

Shen-Yun Miao, Chao-Chun Liang, Keh-Yih Su

This paper reviews the evolution of pre-trained models (PTMs), focusing on architectures like BERT and GPT, highlighting their transformative impact on AI.

Xu Han, Zhengyan Zhang, Ning Ding et al.

Proposes Traffic Transformer, using multi-head attention to dynamically learn spatial-temporal traffic features, greatly improving long-term prediction accuracy.

Haoyang Yan, Xiaolei Ma

Proposes a Patternist-based AGI architecture emphasizing cognitive synergy, using Typed Metagraphs and Galois connections for knowledge and process integration.

Ben Goertzel

This paper systematically introduces multi-objective reinforcement learning and planning, emphasizing trade-offs in complex decision-making, with innovative algorithms and real-world applications.

Conor F. Hayes, Roxana Rădulescu, Eugenio Bargiacchi et al.

Applying GPT-3 for email understanding and response automation, demonstrating technical and economic viability to enhance productivity.

Jonas Thiergart, Stefan Huber, Thomas Übellacker

Proposes 12 new definitions of actual causation by integrating causal sufficiency and necessity, validated through structural equation models and classic examples.

Sander Beckers

This study evaluates LIME, SHAP, and TreeInterpreter in real-world fraud detection, revealing explanations can sometimes reduce decision accuracy.

Sérgio Jesus, Catarina Belém, Vladimir Balayan et al.

TENSAT employs equality saturation with e-graphs for tensor graph superoptimization, achieving up to 16% speedup and 48x faster optimization.

Yichen Yang, Phitchaya Mangpo Phothilimtha, Yisu Remy Wang et al.

Proposes MARL, combining weakly supervised retrieval with meta-learning, achieving SOTA (micro F177.71%) in KBQA

Yuncheng Hua, Yuan-Fang Li, Gholamreza Haffari et al.

EECBS integrates explicit estimation search and online learning to accelerate bounded-suboptimal multi-agent pathfinding, outperforming state-of-the-art methods.

Jiaoyang Li, Wheeler Ruml, Sven Koenig