Search

At the risk of overstating the case, connectionist approaches to machine learning, i.e. neural networks, are enjoying a small vogue right now. However, these methods require large volumes of data and produce models that are uninterpretable to humans. An alternative framework that is compatible with neural networks and gradient-based learning, but explicitly models compositionality, is Vector Symbolic Architectures (VSAs). VSAs are a family of algebras on high-dimensional vector representations. They arose in cognitive science from the need to unify neural processing and the kind of symbolic reasoning that humans perform. While machine learning methods have benefited from category theoretical analyses, VSAs have not yet received similar treatment. In this paper, we present a first attempt at applying category theory to VSAs. Specifically, we conduct a brief literature survey demonstrating the lacking intersection of these two topics, provide a list of desiderata for VSAs, and propose that VSAs may be understood as a (division) rig in a category enriched over a monoid in Met (the category of Lawvere metric spaces). This final contribution suggests that VSAs may be generalised beyond current implementations. It is our hope that grounding VSAs in category theory will lead to more rigorous connections with other research, both within and beyond, learning and cognition., Comment: 13 pages, no figures, 2 tables, one appendix

Neuromorphic computing shows promise for advancing computing efficiency and capabilities of AI applications using brain-inspired principles. However, the neuromorphic research field currently lacks standardized benchmarks, making it difficult to accurately measure technological advancements, compare performance with conventional methods, and identify promising future research directions. Prior neuromorphic computing benchmark efforts have not seen widespread adoption due to a lack of inclusive, actionable, and iterative benchmark design and guidelines. To address these shortcomings, we present NeuroBench: a benchmark framework for neuromorphic computing algorithms and systems. NeuroBench is a collaboratively-designed effort from an open community of researchers across industry and academia, aiming to provide a representative structure for standardizing the evaluation of neuromorphic approaches. The NeuroBench framework introduces a common set of tools and systematic methodology for inclusive benchmark measurement, delivering an objective reference framework for quantifying neuromorphic approaches in both hardware-independent (algorithm track) and hardware-dependent (system track) settings. In this article, we outline tasks and guidelines for benchmarks across multiple application domains, and present initial performance baselines across neuromorphic and conventional approaches for both benchmark tracks. NeuroBench is intended to continually expand its benchmarks and features to foster and track the progress made by the research community., Comment: To appear in Nature Neuromorphic Hardware and Computing collection

A close partnership between people and partially autonomous machines has enabled decades of space exploration. But to further expand our horizons, our systems must become more capable. Increasing the nature and degree of autonomy - allowing our systems to make and act on their own decisions as directed by mission teams - enables new science capabilities and enhances science return. The 2011 Planetary Science Decadal Survey (PSDS) and on-going pre-Decadal mission studies have identified increased autonomy as a core technology required for future missions. However, even as scientific discovery has necessitated the development of autonomous systems and past flight demonstrations have been successful, institutional barriers have limited its maturation and infusion on existing planetary missions. Consequently, the authors and endorsers of this paper recommend that new programmatic pathways be developed to infuse autonomy, infrastructure for support autonomous systems be invested in, new practices be adopted, and the cost-saving value of autonomy for operations be studied., Comment: 10 pages (compared to 8 submitted to PSADS), 2 figures, submitted to National Academy of Sciences Planetary Science and Astrobiology Decadal Survey 2023-2032

Robotic science missions in remote environments, such as deep ocean and outer space, can involve studying phenomena that cannot directly be observed using on-board sensors but must be deduced by combining measurements of correlated variables with domain knowledge. Traditionally, in such missions, robots passively gather data along prescribed paths, while inference, path planning, and other high level decision making is largely performed by a supervisory science team. However, communication constraints hinder these processes, and hence the rate of scientific progress. This paper presents an active perception approach that aims to reduce robots' reliance on human supervision and improve science productivity by encoding scientists' domain knowledge and decision making process on-board. We use Bayesian networks to compactly model critical aspects of scientific knowledge while remaining robust to observation and modeling uncertainty. We then formulate path planning and sensor scheduling as an information gain maximization problem, and propose a sampling-based solution based on Monte Carlo tree search to plan informative sensing actions which exploit the knowledge encoded in the network. The computational complexity of our framework does not grow with the number of observations taken and allows long horizon planning in an anytime manner, making it highly applicable to field robotics. Simulation results show statistically significant performance improvements over baseline methods, and we validate the practicality of our approach through both hardware experiments and simulated experiments with field data gathered during the NASA Mojave Volatiles Prospector science expedition.

Our neural architecture search algorithm progressively searches a tree of neural network architectures. Child nodes are created by inserting new layers determined by a transition graph into a parent network up to a maximum depth and pruned when performance is worse than its parent. This increases efficiency but makes the algorithm greedy. Simpler networks are successfully found before more complex ones that can achieve benchmark performance similar to other top-performing networks.

This paper addresses the problem of localization for landing on the surface of icy moons, like Europa or Enceladus. Due to the possibility of specular reflection as well as high bulk albedo, icy surfaces present new challenges that make traditional vision-based navigation systems relying on visible imagery unreliable. We propose augmenting visible light cameras with a thermal-infrared camera using inverse-depth parameterized monocular EKF-SLAM to address problems arising from the appearance of icy moons. Results were obtained from a novel procedural Europa surface simulation which models the appearance and the thermal properties simultaneously from physically-based methods. In this framework, we show that thermal features improve localization by 23% on average when compared to a visible camera. Moreover, fusing both sensing modalities increases the improvement in localization to 31% on average, compared to using a visible light camera alone.

In robotic information gathering missions, scientists are typically interested in understanding variables which require proxy measurements from specialized sensor suites to estimate. However, energy and time constraints limit how often these sensors can be used in a mission. Robots are also equipped with cheaper to use navigation sensors such as cameras. In this paper, we explore a challenging planning problem in which a robot is required to learn about a scientific variable of interest in an initially unknown environment by planning informative paths and deciding when and where to use its sensors. To tackle this we present two innovations: a Bayesian generative model framework to automatically learn correlations between expensive science sensors and cheaper to use navigation sensors online, and a sampling based approach to plan for multiple sensors while handling long horizons and budget constraints. Our approach does not grow in complexity with data and is anytime making it highly applicable to field robotics. We tested our approach extensively in simulation and validated it with real data collected during the 2014 Mojave Volatiles Prospector Mission. Our planning algorithm performs statistically significantly better than myopic approaches and at least as well as a coverage-based algorithm in an initially unknown environment while having added advantages of being able to exploit prior knowledge and handle other intricacies of the real world without further algorithmic modifications.

This paper presents a new algorithm for autonomous on-line exploration in unknown environments. The objective of the algorithm is to free robot scientists from extensive preliminary site investigation while still being able to collect meaningful data. We simulate a common form of exploration task for an autonomous robot involving sampling the environment at various locations and compare performance with a simpler existing algorithm that is also denied global information. The result of the experiment shows that the new algorithm has a statistically significant improvement in performance with a significant effect size for a range of costs for taking sampling actions.

This paper presents a new algorithm for autonomous on-line exploration in unknown environments. The objective is to free remote scientists from possibly-infeasible extensive preliminary site investigation prior to sending robotic agents. We simulate a common exploration task for an autonomous robot sampling the environment at various locations and compare performance against simpler control strategies. An extension is proposed and evaluated that further permits operation in the presence of environmental variability in which the robot encounters a change in the distribution underlying sampling targets. Experimental results indicate a strong improvement in performance across varied parameter choices for the scenario.