Efficient Training of Artificial Neural Networks for Autonomous Navigation

By Dean A. Pomerleau. 1991

Subfields: Reinforcement Learning, Model Architectures

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Reinforcement Learning

In Model Architectures

Citations in the corpus

Scalable agent alignment via reward modeling: a research direction (2018, background): “One strategy to train aligned agents could be from imitating human behavior (Pomerleau, 1991; Abbeel & Ng, 2004; Ho & Ermon, 2016; Finn et al., 2016).”
Pre-Trained Language Models for Interactive Decision-Making (2022, background): “behavior cloning (Pomerleau, 1991; 1989; Torabi et al., 2018), to train the policy.”
Generative Adversarial Imitation Learning (2016, background): “There are two main approaches suitable for this setting: behavioral cloning [20], which learns a policy as a supervised learning problem over state-action pairs from expert trajectories; and inverse reinforcement learning [25, 18], which ﬁnds a cost function under which the expert is uniquely…”
Time-Contrastive Networks: Self-Supervised Learning from Video (2017, background): “where examples of behaviors are provided as state-action pairs [8, 9].”
Deep Reinforcement Learning for Autonomous Driving: A Survey (2020, background): “[112], [113] presented agents that learn form demonstrations (LfD) that tries to mimic the behavior of an expert.”
A Minimalist Approach to Offline Reinforcement Learning (2021, background): “Behavior cloning (BC) is an approach for imitation learning [Pomerleau, 1991], where the policy is trained with supervised learning to directly imitate the actions of a provided dataset.”
Continuous Doubly Constrained Batch Reinforcement Learning (2021, methodology): “In fact, this is the well-known likelihood based behavioral cloning algorithm used by Pomerleau (1991).”
Provable Representation Learning for Imitation with Contrastive Fourier Features (2021, methodology): “This focus puts us closer to the supervised learning regime, and, indeed, a common approach to imitation learning – known as behavioral cloning (BC) – is to perform max-likelihood training on a collected a set of target demonstrations composed of state-action pairs sampled from the target policy [31”
Hyperparameter Selection for Imitation Learning (2021, methodology): “IL can be achieved through pure supervised learning (Pomerleau, 1991) but many IL approaches leverage the assumption that *Equal contribution 1Google Research, Brain Team 2Univ. de Lille, CNRS, Inria Scool, UMR 9189 CRIStAL.”
Representation Matters: Offline Pretraining for Sequential Decision Making (2021, methodology): “To mimic this practical scenario, we consider an experimental protocol in which the downstream task is behavioral cloning (Pomerleau, 1991) on a small set of expert trajectories – selected by taking either the ﬁrst 10k or 25k transitions from an expert dataset in D4RL, corresponding to about 10 and…”
Robust Imitation of Diverse Behaviors (2017, background): “These limitations of supervised learning for imitation, also known as behavioral cloning (BC) [24], are well known [27, 28].”
Hippocampal Replay is Not a Simple Function of Experience (2010, background): “…replaying nonlocal trajectories when they are not being visited (as was the case for left-only and right-only half-sessions but not for alternation half-sessions) could be a mechanism to prevent the representation of the nonrecently experienced trajectories from degrading (Pomerleau, 1991).”
Model-Free Imitation Learning with Policy Optimization (2016, background): “An alternative and often more practical approach, called imitation learning , is to encode preferences and differentiate between desirable and undesirable outcomes using demonstrations provided by an expert (Pomerleau, 1991; Russell, 1998).”
Extrapolating Beyond Suboptimal Demonstrations via Inverse Reinforcement Learning from Observations (2019, background): “Early work on learning from demonstration focused on behavioral cloning (Pomerleau, 1991), in which the goal is to learn a policy that imitates the actions taken by the demonstrator; however, without substantial human feedback and correction, this method is known to have large generalization error (”
A Bayesian Approach to Generative Adversarial Imitation Learning (2018, background): “Historically, behavioral cloning (BC) [Pomerleau, 1991] is one of the simplest approach to imitation learning, which learns to map the states to demonstrated actions using supervised learning.”
Energy-Based Imitation Learning (2020, methodology): “Thus, Behavior Cloning (BC) [32] provides a straightforward method by”