Guiding generative AI
Related Articles
Preserving and combining knowledge in robotic lifelong reinforcement learning
Humans can continually accumulate knowledge and develop increasingly complex behaviours and skills throughout their lives, which is a capability known as ‘lifelong learning’. Although this lifelong learning capability is considered an essential mechanism that makes up general intelligence, recent advancements in artificial intelligence predominantly excel in narrow, specialized domains and generally lack this lifelong learning capability. Here we introduce a robotic lifelong reinforcement learning framework that addresses this gap by developing a knowledge space inspired by the Bayesian non-parametric domain. In addition, we enhance the agent’s semantic understanding of tasks by integrating language embeddings into the framework. Our proposed embodied agent can consistently accumulate knowledge from a continuous stream of one-time feeding tasks. Furthermore, our agent can tackle challenging real-world long-horizon tasks by combining and reapplying its acquired knowledge from the original tasks stream. The proposed framework advances our understanding of the robotic lifelong learning process and may inspire the development of more broadly applicable intelligence.
Categorizing robots by performance fitness into the tree of robots
Robots are typically classified based on specific morphological features, like their kinematic structure. However, a complex interplay between morphology and intelligence shapes how well a robot performs processes. Just as delicate surgical procedures demand high dexterity and tactile precision, manual warehouse or construction work requires strength and endurance. These process requirements necessitate robot systems that provide a level of performance fitting the process. In this work, we introduce the tree of robots as a taxonomy to bridge the gap between morphological classification and process-based performance. It classifies robots based on their fitness to perform, for example, physical interaction processes. Using 11 industrial manipulators, we constructed the first part of the tree of robots based on a carefully deduced set of metrics reflecting fundamental robot capabilities for various industrial physical interaction processes. Through significance analysis, we identified substantial differences between the systems, grouping them via an expectation-maximization algorithm to create a fitness-based robot classification that is open for contributions and accessible.
Two types of motifs enhance human recall and generalization of long sequences
Whether it is listening to a piece of music, learning a new language, or solving a mathematical equation, people often acquire abstract notions in the sense of motifs and variables—manifested in musical themes, grammatical categories, or mathematical symbols. How do we create abstract representations of sequences? Are these abstract representations useful for memory recall? In addition to learning transition probabilities, chunking, and tracking ordinal positions, we propose that humans also use abstractions to arrive at efficient representations of sequences. We propose and study two abstraction categories: projectional motifs and variable motifs. Projectional motifs find a common theme underlying distinct sequence instances. Variable motifs contain symbols representing sequence entities that can change. In two sequence recall experiments, we train participants to remember sequences with projectional and variable motifs, respectively, and examine whether motif training benefits the recall of novel sequences sharing the same motif. Our result suggests that training projectional and variables motifs improve transfer recall accuracy, relative to control groups. We show that a model that chunks sequences in an abstract motif space may learn and transfer more efficiently, compared to models that learn chunks or associations on a superficial level. Our study suggests that humans construct efficient sequential memory representations according to the two types of abstraction we propose, and creating these abstractions benefits learning and out-of-distribution generalization. Our study paves the way for a deeper understanding of human abstraction learning and generalization.
Modelling and design of transcriptional enhancers
Transcriptional enhancers are the genomic elements that contain critical information for the regulation of gene expression. This information is encoded through precisely arranged transcription factor-binding sites. Genomic sequence-to-function models, computational models that take DNA sequences as input and predict gene regulatory features, have become essential for unravelling the complex combinatorial rules that govern cell-type-specific activities of enhancers. These models function as biological ‘oracles’, capable of accurately predicting the activity of novel DNA sequences. By leveraging these oracles, DNA sequences can be optimized towards designed synthetic enhancers with tailored cell-type-specific or cell-state-specific activities. In parallel, generative artificial intelligence is rapidly advancing in genomics and enhancer design. Synthetic enhancers hold great promise for a wide range of biomedical applications, from facilitating fundamental research to enabling gene therapies.
Modeling the impact of structure and coverage on the reactivity of realistic heterogeneous catalysts
Adsorbates often cover the surfaces of catalysts densely as they carry out reactions, dynamically altering their structure and reactivity. Understanding adsorbate-induced phenomena and harnessing them in our broader quest for improved catalysts is a substantial challenge that is only beginning to be addressed. Here we chart a path toward a deeper understanding of such phenomena by focusing on emerging in silico modeling methodologies, which will increasingly incorporate machine learning techniques. We first examine how adsorption on catalyst surfaces can lead to local and even global structural changes spanning entire nanoparticles, and how this affects their reactivity. We then evaluate current efforts and the remaining challenges in developing robust and predictive simulations for modeling such behavior. Last, we provide our perspectives in four critical areas—integration of artificial intelligence, building robust catalysis informatics infrastructure, synergism with experimental characterization, and adaptive modeling frameworks—that we believe can help surmount the remaining challenges in rationally designing catalysts in light of these complex phenomena.
Responses