We present a new approach for efficient exploration which leverages a low-dimensional encoding of the environment learned with a combination of model-based and model-free objectives. Our approach uses intrinsic rewards that are based on the distance of nearest neighbors in the low dimensional representational space to gauge novelty.

The nouns of our language refer to either concrete entities (like a table) or abstract concepts (like justice or love), and cognitive psychology has established that concreteness influences how words are processed. Accordingly, understanding how concreteness is represented in our mind and brain is a central question in psychology, neuroscience, and computational linguistics. While the advent of powerful language models has allowed for quantitative inquiries into the nature of semantic representations, it remains largely underexplored how they represent concreteness. Here, we used behavioral judgments to estimate semantic distances implicitly used by humans, for a set of carefully selected abstract and concrete nouns. Using Representational Similarity Analysis, we find that the implicit representational space of participants and the semantic representations of language models are significantly aligned. We also find that both representational spaces are implicitly aligned to an explicit representation of concreteness, which was obtained from our participants using an additional concreteness rating task. Importantly, using ablation experiments, we demonstrate that the human-to-model alignment is substantially driven by concreteness, but not by other important word characteristics established in psycholinguistics. These results indicate that humans and language models converge on the concreteness dimension, but not on other dimensions.

Additional Feedback: The method seems to be restricted to deterministic environments. Could we add a bit of discussion why it would be the case and how we could imagine to extend the approach to deal with stochastic environments (maybe in the supplementary material)? In most approaches, the discount factor is an exponential function of the distance in time, why did the authors choose to make it a function of state and action, and why should we learn it? Having the environment return the discount factor is not really common. The choice of the learned representation size seems to contain some domain knowledge.

This paper proposes an novelty-search exploration method based on an encoding of the environment. Their method computes the novelty of a state in a learned representation embedding space and encourages the agent to optimize for this novelty using a combined model-free and model-based approach. Motivated by the information bottleneck principle, the embedding space is learned by maximizing compression while retaining an accurate dynamics model, resulting in compressing the environment into a small state space well-suited for novelty-based exploration. The experiments were also clear and well-motivated, on grid-type domains to evaluate state coverage, and also two control domains to evaluate the improvement of novelty search on the agent's ability to perform control tasks. I particularly enjoyed the learned abstract visualization of the labyrinth env in Figure 1.

We demonstrate the successful use of scattering representations without further compression for simulation-based inference (SBI) with images (i.e. field-level), illustrated with a cosmological case study. Scattering representations provide a highly effective representational space for subsequent learning tasks, although the higher dimensional compressed space introduces challenges. We overcome these through spatial averaging, coupled with more expressive density estimators. Compared to alternative methods, such an approach does not require additional simulations for either training or computing derivatives, is interpretable, and resilient to covariate shift. As expected, we show that a scattering only approach extracts more information than traditional second order summary statistics.

We present a new approach for efficient exploration which leverages a low-dimensional encoding of the environment learned with a combination of model-based and model-free objectives. Our approach uses intrinsic rewards that are based on the distance of nearest neighbors in the low dimensional representational space to gauge novelty. One key element of our approach is the use of information theoretic principles to shape our representations in a way so that our novelty reward goes beyond pixel similarity. We test our approach on a number of maze tasks, as well as a control problem and show that our exploration approach is more sample-efficient compared to strong baselines.