Search for: All records

Electrification of buildings through deployment of heat pumps requires innovative design and control strategies to reduce their energy demands on the grid. Instead of the sequential approach of optimizing the design specifications and control strategies, this paper considers the benefits of the combined and simultaneous optimization of design capacities and control settings for heat pumps when specified for US residential buildings. A Genetic Algorithm optimizer is used to simultaneously adjust the main and supplementary coil capacities for the heat pump as well as the indoor temperature setpoints to minimize annual heating and cooling energy needs as well as occupant thermal discomfort levels. In comparison to design and control baselines, it is found that simultaneous optimization can achieve 21% and 7% reductions in heating and cooling annual energy consumption for the cases of variable speed and single speed heat pumps. Moreover, the analysis results indicate that these reductions are nearly double the savings obtained when design only and control only based optimizations are considered. The presented combined design and control optimization approach could potentially provide an effective paradigm shift in specifying heat pump systems for residential buildings. 

The nexus between technology and workplace inequality has been a long-standing topic of scholarly interest, now heightened by the rapid evolution of artificial intelligence (AI). Our review moves beyond dystopian or utopian views of AI by identifying four perspectives—normative, cognitive, structural, and relational—espoused by scholars examining the impact of AI on workplace inequality specifically, and the structure and organization of work more broadly. We discuss the respective strengths, limitations, and underlying assumptions of these perspectives and highlight how each perspective speaks to a particular facet of workplace inequality: either encoded, evaluative, wage, or relational inequality. Integrating these perspectives enables a deeper understanding of the mechanisms, processes, and trajectories through which AI influences workplace inequality, as well as the role that organizational managers, workers, and policymakers could play in the process. Toward this end, we introduce a framework on the “inequality cascades” of AI that traces how and when inequality emerges and amplifies cumulatively as AI systems progress through the phases of development, implementation, and use in organizations. In turn, we articulate a research agenda for management and organizational scholars to better understand AI and its multifaceted impact on workplace inequality, and we examine potential mechanisms to mitigate its adverse consequences. 

All biological systems are subject to perturbations arising from thermal fluctuations, external environments, or mutations. Yet, while biological systems consist of thousands of interacting components, recent high-throughput experiments have shown that their response to perturbations is surprisingly low dimensional: confined to only a few stereotyped changes out of the many possible. In this review, we explore a unifying dynamical systems framework—soft modes—to explain and analyze low dimensionality in biology, from molecules to ecosystems. We argue that this soft mode framework makes nontrivial predictions that generalize classic ideas from developmental biology to disparate systems, namely phenocopying, dual buffering, and global epistasis. While some of these predictions have been borne out in experiments, we discuss how soft modes allow for a surprisingly far-reaching and unifying framework in which to analyze data from protein biophysics to microbial ecology. 

Representation Learning), a novel multimodal meta-learning framework for few-shot learning in heterogeneous systems, designed for science and engineering problems where entities share a common underlying forward model but exhibit heterogeneity due to entity-specific characteristics. TAM-RL leverages an amortized training process with a modulation network and a base network to learn task-specific modulation parameters, enabling efficient adaptation to new tasks with limited data. We evaluate TAM-RL on two real-world environmental datasets: Gross Primary Product (GPP) prediction and streamflow forecasting, demonstrating significant improvements over existing meta-learning methods. On the FLUXNET dataset, TAM-RL improves RMSE by 18.9% over MMAML with just one month of few-shot data, while for streamflow prediction, it achieves an 8.21% improvement with one year of data. Synthetic data experiments further validate TAM-RL’s superior performance in heterogeneous task distributions, outperforming the baselines in the most heterogeneous setting. Notably, TAM-RL offers substantial computational efficiency, with at least 3x faster training times compared to gradient-based meta-learning approaches while being much simpler to train due to reduced complexity. Ablation studies highlight the importance of pretraining and adaptation mechanisms in TAM-RL’s performance. Keywords: Representation Learning, meta-learning, few-shot learning, environmental applications, time-series. DOI:10.1137/1.9781611978520.2