skip to main content

Search for: All records

Creators/Authors contains: "Kumar A"

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. A popular line of recent research incorporates ML advice in the design of online algorithms to improve their performance in typical instances. These papers treat the ML algorithm as a blackbox, and redesign online algorithms to take advantage of ML predictions. In this paper, we ask the complementary question: can we redesign ML algorithms to provide better predictions for online algorithms? We explore this question in the context of the classic rent-or-buy problem, and show that incorporating optimization benchmarks directly in ML loss functions leads to significantly better performance, while maintaining a worst-case adversarial result when the advice is completely wrong. We support this finding both through theoretical bounds and numerical simulations, and posit that “learning for optimization” is a fertile area for future research.
    Free, publicly-accessible full text available July 1, 2023
  2. Free, publicly-accessible full text available July 1, 2023
  3. Human-AI collaboration is an increasingly commonplace part of decision-making in real world applications. However, how humans behave when collaborating with AI is not well understood. We develop metacognitive bandits, a computational model of a human's advice-seeking behavior when working with an AI. The model describes a person's metacognitive process of deciding when to rely on their own judgment and when to solicit the advice of the AI. It also accounts for the difficulty of each trial in making the decision to solicit advice. We illustrate that the metacognitive bandit makes decisions similar to humans in a behavioral experiment. We also demonstrate that algorithm aversion, a widely reported bias, can be explained as the result of a quasi-optimal sequential decision-making process. Our model does not need to assume any prior biases towards AI to produce this behavior.
  4. East Antarctica is covered by thick sheets of ice and is underlain by stable cratonic lithosphere, extensive mountain ranges, and subglacial basins. The sparse seismic coverage in this region makes it difficult to assess the crustal and mantle structure, which are important to understanding the tectonic evolution of the continent as well as the behavior of the overlying ice sheets. Present tomographic models lack resolution and are often inconsistent with one another; therefore, delineating sub-surface characteristics associated with old rift systems or structures that would allow us to assess the origins of the Wilkes and Aurora subglacial basins, for instance, becomes challenging. To overcome these limitations, we are using a full-waveform tomography method to model the crustal and upper mantle structure in East Antarctica. We have used a frequency-time normalization approach to extract empirical Green’s functions (EGFs) from ambient seismic noise, between periods of 15-340 seconds. The ray path coverage of the EGFs is dense throughout East Antarctica, indicating that our study will provide new, high resolution imaging of this area. Synthetic waveforms are simulated through a three-dimensional heterogeneous Earth model using a finite-difference wave propagation method with a grid spacing of 0.025º (~ 2.25 km), which accurately reproduce Rayleighmore »waves at 15+ seconds. Following this, phase delays are measured between the synthetics and the data, sensitivity kernels are constructed using a scattering integral approach, and we invert using a sparse, least-squares method. The resulting shear-wave velocity model will be used to assess crustal and upper mantle features, ultimately aimed at resolving whether old rift systems exist within East Antarctica in relation to prominent subglacial basins. Preliminary results will be shared.« less