Эффективность выборки
Related Papers
Efficient Meta Lifelong-Learning with Limited Memory
AbstractCurrent natural language processing models work well on a single task, yet they often fail to continuously learn new tasks without forgetting previous ones as they are re-trained throughout their …...
Mind Your Outliers! Investigating the Negative Impact of Outliers on Active Learning for Visual Question Answering
AbstractActive learning promises to alleviate the massive data needs of supervised machine learning: it has successfully improved sample efficiency by an order of magnitude on traditional tasks like topic classification …...
Multitasking Inhibits Semantic Drift
AbstractWhen intelligent agents communicate to accomplish shared goals, how do these goals shape the agents’ language? We study the dynamics of learning in latent language policies (LLPs), in which instructor …...
[CASPI] Causal-aware Safe Policy Improvement for Task-oriented Dialogue
AbstractThe recent success of reinforcement learning (RL) in solving complex tasks is often attributed to its capacity to explore and exploit an environment. Sample efficiency is usually not an issue …...
Data-to-text Generation with Variational Sequential Planning
AbstractWe consider the task of data-to-text generation, which aims to create textual output from non-linguistic input. We focus on generating long-form text, that is, documents with multiple paragraphs, and propose …...
Perceiving the World: Question-guided Reinforcement Learning for Text-based Games
AbstractText-based games provide an interactive way to study natural language processing. While deep reinforcement learning has shown effectiveness in developing the game playing agent, the low sample efficiency and the …...
Platt-Bin: Efficient Posterior Calibrated Training for NLP Classifiers
AbstractModern NLP classifiers are known to return uncalibrated estimations of class posteriors. Existing methods for posterior calibration rescale the predicted probabilities but often have an adverse impact on final classification …...
Combining Parameter-efficient Modules for Task-level Generalisation
AbstractA modular design encourages neural models to disentangle and recombine different facets of knowledge to generalise more systematically to new tasks. In this work, we assume that each task is …...
FedPerC: Federated Learning for Language Generation with Personal and Context Preference Embeddings
AbstractFederated learning is a training paradigm that learns from multiple distributed users without aggregating data on a centralized server, promising the ability to deploy machine-learning to a diverse population of …...