Task Embeddings

Task embeddings are low-dimensional vector representations designed to capture the essential characteristics of different machine learning tasks. Current research focuses on improving the accuracy and consistency of these embeddings, particularly for predicting task transferability in multi-task learning scenarios, and exploring their use across diverse model architectures, including transformers and graph neural networks. This work is significant because effective task embeddings can enhance the efficiency of model training and adaptation, leading to improved performance in various applications such as natural language processing, robotics, and anomaly detection. The development of unified task embedding frameworks that bridge the gap between different model types is a particularly active area of investigation.

Papers

July 23, 2024

Exploring the Effectiveness and Consistency of Task Selection in Intermediate-Task Transfer Learning
Pin-Jie Lin, Miaoran Zhang, Marius Mosbach, Dietrich Klakow
Transfer Learning Strong Consistency Task Transferability Task Embeddings Task Selection

March 1, 2024

Task Indicating Transformer for Task-conditional Dense Predictions
Yuxiang Lu, Shalayiding Sirejiding, Bayram Bayramli, Suizhi Huang, Yue Ding, Hongtao Lu
Transformer Based Multi Task Learning Related Task Dense Prediction Task Embeddings Task Specific Representation Multi Task Transfer

February 22, 2024

Towards Unified Task Embeddings Across Multiple Models: Bridging the Gap for Prompt-Based Large Language Models and Beyond
Xinyu Wang, Hainiu Xu, Lin Gui, Yulan He
Large Language Model Multi Task Learning Multiple Model Task Embeddings Task Specific Language Model

February 12, 2024

Task-conditioned adaptation of visual features in multi-task policy learning
Pierre Marza, Laetitia Matignon, Olivier Simonin, Christian Wolf
Vision Model Visual Feature Task Adaptation Unseen Task Task Specific Adapter Task Embeddings Multi Task Policy

June 5, 2023

Learning Embeddings for Sequential Tasks Using Population of Agents
Mridul Mahajan, Georgios Tzannetos, Goran Radanovic, Adish Singla
Reinforcement Learning Agent Smith Na\"ive Population Learned Embeddings Information Theoretic Framework Sequential Task Task Embeddings

March 14, 2023

AutoTransfer: AutoML with Knowledge Transfer -- An Application to Graph Neural Networks
Kaidi Cao, Jiaxuan You, Jiaju Liu, Jure Leskovec
Graph Neural Network Application Proficiency Knowledge Transfer Online autoML GNN Architecture Task Model AutoML Approach Task Embeddings

October 21, 2022

Efficiently Tuned Parameters are Task Embeddings
Wangchunshu Zhou, Canwen Xu, Julian McAuley
Many Parameter Task Similarity Model Transferability Task Embeddings

October 10, 2022

Using Both Demonstrations and Language Instructions to Efficiently Learn Robotic Tasks
Albert Yu, Raymond J. Mooney
Noisy Demonstration Robotic Task Language Conditioned Novel Task Task Specific Policy Task Embeddings Language Instruction

September 20, 2022

Collaborative Anomaly Detection
Ke Bai, Aonan Zhang, Zhizhong Li, Ricardo Heano, Chong Wang, Lawrence Carin
Anomaly Detection Multi Task Learning Single Task Learning Task Embeddings Collaborative Anomaly Detection

July 5, 2022

Learning Task Embeddings for Teamwork Adaptation in Multi-Agent Reinforcement Learning
Lukas Schäfer, Filippos Christianos, Amos Storkey, Stefano V. Albrecht
Multi Agent Reinforcement Learning Unknown Teammate Task Embeddings

April 29, 2022

Task Embedding Temporal Convolution Networks for Transfer Learning Problems in Renewable Power Time-Series Forecast
Jens Schreiber, Stephan Vogt, Bernhard Sick
Multi Task Learning Related Task Temporal Convolutional Network Renewable Energy Task Embeddings Solar Data

April 12, 2022

Arch-Graph: Acyclic Architecture Relation Predictor for Task-Transferable Neural Architecture Search
Minbin Huang, Zhijian Huang, Changlin Li, Xin Chen, Hang Xu, Zhenguo Li, Xiaodan Liang
Neural Architecture Search Relation Prediction Task Embeddings Optimal Neural Architecture Graph Architecture

December 10, 2021

Analysis and Prediction of NLP Models Via Task Embeddings
Damien Sileo, Marie-Francine Moens
Zero Shot General Analysis Human Prediction Jina Embeddings NLP Task Task Embeddings Task Engagement

Task Embeddings

Papers

Exploring the Effectiveness and Consistency of Task Selection in Intermediate-Task Transfer Learning

Task Indicating Transformer for Task-conditional Dense Predictions

Towards Unified Task Embeddings Across Multiple Models: Bridging the Gap for Prompt-Based Large Language Models and Beyond

Task-conditioned adaptation of visual features in multi-task policy learning

Learning Embeddings for Sequential Tasks Using Population of Agents

AutoTransfer: AutoML with Knowledge Transfer -- An Application to Graph Neural Networks

Efficiently Tuned Parameters are Task Embeddings

Using Both Demonstrations and Language Instructions to Efficiently Learn Robotic Tasks

Collaborative Anomaly Detection

Learning Task Embeddings for Teamwork Adaptation in Multi-Agent Reinforcement Learning

Task Embedding Temporal Convolution Networks for Transfer Learning Problems in Renewable Power Time-Series Forecast

Arch-Graph: Acyclic Architecture Relation Predictor for Task-Transferable Neural Architecture Search

Analysis and Prediction of NLP Models Via Task Embeddings