Search for: All records

Two major barriers hinder the holistic understanding of subsurface critical zone (CZ) evolution and its impacts: (a) an inability to measure, define, and share information and (b) a societal structure that inhibits inclusivity and creativity. In contrast to the aboveground portion of the CZ, which is visible and measurable, the bottom boundary is difficult to access and quantify. In the context of these barriers, we aim to expand the spatial reach of the CZ by highlighting existing and effective tools for research as well as the “human reach” of CZ science by expanding who performs such science and who it benefits. We do so by exploring the diversity of vocabularies and techniques used in relevant disciplines, defining terminology, and prioritizing research questions that can be addressed. Specifically, we explore geochemical, geomorphological, geophysical, and ecological measurements and modeling tools to estimate CZ base and thickness. We also outline the importance of and approaches to developing a diverse CZ workforce that looks like and harnesses the creativity of the society it serves, addressing historical legacies of exclusion. Looking forward, we suggest that to grow CZ science, we must broaden the physical spaces studied and their relationships with inhabitants, measure the “deep” CZ and make data accessible, and address the bottlenecks of scaling and data‐model integration. What is needed—and what we have tried to outline—are common and fundamental structures that can be applied anywhere and used by the diversity of researchers involved in investigating and recording CZ processes from a myriad of perspectives.

Embedding is widely used in recommendation models to learn feature representations. However, the traditional embedding technique that assigns a fixed size to all categorical features may be suboptimal due to the following reasons. In recommendation domain, the majority of categorical features' embeddings can be trained with less capacity without impacting model performance, thereby storing embeddings with equal length may incur unnecessary memory usage. Existing work that tries to allocate customized sizes for each feature usually either simply scales the embedding size with feature's popularity or formulates this size allocation problem as an architecture selection problem. Unfortunately, most of these methods either have large performance drop or incur significant extra time cost for searching proper embedding sizes. In this article, instead of formulating the size allocation problem as an architecture selection problem, we approach the problem from a pruning perspective and proposePruning-basedMulti-sizeEmbedding (PME) framework. During the search phase, we prune the dimensions that have the least impact on model performance in the embedding to reduce its capacity. Then, we show that the customized size of each token can be obtained by transferring the capacity of its pruned embedding with significant less search cost. Experimental results validate that PME can efficiently find proper sizes and hence achieve strong performance while significantly reducing the number of parameters in the embedding layer.