Publications

While the vision of accelerating materials discovery using data driven methods is well-founded, practical realization has been throttled due to challenges in data generation, ingestion, and materials state-aware machine learning. High-throughput experiments and automated computational workflows are addressing the challenge of data generation, and capitalizing on these emerging data resources requires ingestion of data into an architecture that captures the complex provenance of experiments and simulations. In this manuscript, we describe an event-sourced architecture for materials provenance (ESAMP) that encodes the sequence and interrelationships among events occurring in a simulation or experiment. We use this architecture to ingest a large and varied dataset (MEAD) that contains raw data and metadata from millions of materials synthesis and characterization experiments performed using various modalities such as serial, parallel, multi-modal experimentation. Our data architecture tracks the evolution of a material's state, enabling a demonstration of how state-equivalency rules can be used to generate datasets that significantly enhance data-driven materials discovery. Specifically, using state-equivalency rules and parameters associated with state-changing processes in addition to the typically used composition data, we demonstrated marked reduction of uncertainty in prediction of overpotential for oxygen evolution reaction (OER) catalysts. Finally, we discuss the importance of ESAMP architecture in enabling several aspects of accelerated materials discovery such as dynamic workflow design, generation of knowledge graphs, and efficient integration of simulation and experiment. READ MORE

Natural Language Processing (NLP), a cornerstone field within artificial intelligence, has been increasingly utilized in the field of materials science literature. Our study conducts a reproducibility analysis of two pioneering works within this domain: "Machine-learned and codified synthesis parameters of oxide materials" by Kim et al., and "Unsupervised word embeddings capture latent knowledge from materials science literature" by Tshitoyan et al. We aim to comprehend these studies from a reproducibility perspective, acknowledging their significant influence on the field of materials informatics, rather than critiquing them. Our study indicates that both papers offered thorough workflows, tidy and well-documented codebases, and clear guidance for model evaluation. This makes it easier to replicate their results successfully and partially reproduce their findings. In doing so, they set commendable standards for future materials science publications to aspire to. However, our analysis also highlights areas for improvement such as to provide access to training data where copyright restrictions permit, more transparency on model architecture and the training process, and specifications of software dependency versions. We also cross-compare the word embedding models between papers, and find that some key differences in reproducibility and cross-compatibility are attributable to design choices outside the bounds of the models themselves. In summary, our study appreciates the benchmark set by these seminal papers while advocating for further enhancements in research reproducibility practices in the field of NLP for materials science. This balance of understanding and continuous improvement will ultimately propel the intersecting domains of NLP and materials science literature into a future of exciting discoveries. READ MORE

Anticipatory thinking (AT) and design have many commonalities. We identify three challenges for all computational AT systems: representation, generation, and evaluation. We discuss how existing artificial intelligence techniques provide some methods for addressing these, but also fall significantly short. Next, we articulate where AT concepts appear in three computational design paradigms: configuration design, design for resilience, and conceptual design. We close by identifying two promising future directions at the intersection of AT and design: modeling other humans and new interfaces to support human decision-makers. READ MORE

Monocular depth estimation is scale-ambiguous, and thus requires scale supervision to produce metric predictions. Even so, the resulting models will be geometry-specific, with learned scales that cannot be directly transferred across domains. Because of that, recent works focus instead on relative depth, eschewing scale in favor of improved up-to-scale zero-shot transfer. In this work we introduce ZeroDepth, a novel monocular depth estimation framework capable of predicting metric scale for arbitrary test images from different domains and camera parameters. This is achieved by (i) the use of input-level geometric embeddings that enable the network to learn a scale prior over objects; and (ii) decoupling the encoder and decoder stages, via a variational latent representation that is conditioned on single frame information. We evaluated ZeroDepth targeting both outdoor (KITTI, DDAD, nuScenes) and indoor (NYUv2) benchmarks, and achieved a new state-of-the-art in both settings using the same pre-trained model, outperforming methods that train on in-domain data and require test-time scaling to produce metric estimates. READ MORE

Materials knowledge is inherently hierarchical. While high-level descriptors such as composition and structure are valuable for contextualizing materials data, the data must ultimately be considered in the context of its low-level acquisition details. Graph databases offer an opportunity to represent hierarchical relationships among data, organizing semantic relationships into a knowledge graph. Herein, we establish a knowledge graph of materials experiments whose construction encodes the complete provenance of each material sample and its associated experimental data and metadata. Additional relationships among materials and experiments further encode knowledge and facilitate data exploration. We illustrate the Materials Experiment Knowledge Graph (MekG) using several use cases, demonstrating the value of modern graph databases for the enterprise of data-driven materials science. READ MORE

Molecular dynamics simulations are useful tools to screen solid polymer electrolytes with suitable properties applicable to Li-ion batteries. However, due to the vast design space of polymers, it is highly desirable to accelerate the screening by reducing the computational time of ion transport properties from simulations. In this study, we show that with a judicious choice of descriptors we can predict the equilibrium ion transport properties in LiTFSI–homopolymer systems within the first 0.5 ns of the production run of simulations. Specifically, we find that descriptors that include information about the behavior of the system, such as ion clustering and time evolution of ion transport properties, have several advantages over polymer structure-based descriptors, as they encode system (polymer and salt) behavior rather than just the class of polymers and can be computed at any time point during the simulations. These characteristics increase the applicability of our descriptors to a wide range of polymer systems (e.g., copolymers, blend of polymers, salt concentrations, and temperatures) and can be impactful in significantly shortening the discovery pipeline for solid polymer electrolytes. READ MORE

Narrative story generation has gained emerging interest in the field of large language models. The present paper aims to compare stories generated by an LLM only (non-interleaved) with those generated by interleaving human-generated and LLM-generated text (interleaved). The study’s hypothesis is that interleaved stories would perform better than non-interleaved stories. To verify this hypothesis, we conducted two tests with roughly 500 participants each. Participants were asked to rate stories of each type, including an overall score or preference and four facets—logical soundness, plausibility, understandability, and novelty. Our findings indicate that interleaved stories were in fact less preferred than non-interleaved stories. The result has implications for the design and implementation of our story generators. This study contributes new insights into the potential uses and restrictions of interleaved and non-interleaved systems regarding generating narrative stories, which may help to improve the performance of such story generators. READ MORE

Denoising diffusion models trained at web-scale have revolutionized image generation. The application of these tools to engineering design is an intriguing possibility, but is currently limited by their inability to parse and enforce concrete engineering constraints. In this paper, we take a step towards this goal by proposing physics-based guidance, which enables optimization of a performance metric (as predicted by a surrogate model) during the generation process. As a proof-of-concept, we add drag guidance to Stable Diffusion, which allows this tool to generate images of novel vehicles while simultaneously minimizing their predicted drag coefficients. READ MORE

Denoising is intuitively related to projection. Indeed, under the manifold hypothesis, adding random noise is approximately equivalent to orthogonal perturbation. Hence, learning to denoise is approximately learning to project. In this paper, we use this observation to interpret denoising diffusion models as approximate gradient descent applied to the Euclidean distance function. We then provide straight-forward convergence analysis of the DDIM sampler under simple assumptions on the projection error of the denoiser. Finally, we propose a new gradient-estimation sampler, generalizing DDIM using insights from our theoretical results. In as few as 5-10 function evaluations, our sampler achieves state-of-the-art FID scores on pretrained CIFAR-10 and CelebA models and can generate high quality samples on latent diffusion models. READ MORE

Denoising is intuitively related to projection. Indeed, under the manifold hypothesis, adding random noise is approximately equivalent to orthogonal perturbation. Hence, learning to denoise is approximately learning to project. In this paper, we use this observation to reinterpret denoising diffusion models as approximate gradient descent applied to the Euclidean distance function. We then provide straight-forward convergence analysis of the DDIM sampler under simple assumptions on the projection-error of the denoiser. Finally, we propose a new sampler based on two simple modifications to DDIM using insights from our theoretical results. In as few as 5-10 function evaluations, our sampler achieves state-of-the-art FID scores on pretrained CIFAR-10 and CelebA models and can generate high quality samples on latent diffusion models. READ MORE