Search for: All records

The stellar companion to the weak-line T Tauri star DI Tau A was first discovered by the lunar occultation technique in 1989 and was subsequently confirmed by a speckle imaging observation in 1991. It has not been detected since, despite being targeted by five different studies that used a variety of methods and spanned more than 20 yr. Here, we report the serendipitous rediscovery of DI Tau B during our Young Exoplanets Spectroscopic Survey (YESS). Using radial velocity data from YESS spanning 17 yr, new adaptive optics observations from Keck II, and a variety of other data from the literature, we derive a preliminary orbital solution for the system that effectively explains the detection and (almost all of the) non-detection history of DI Tau B. We estimate the dynamical masses of both components, finding that the large mass difference (q ∼ 0.17) and long orbital period (>35 yr) make the DI Tau system a noteworthy and valuable addition to studies of stellar evolution and pre-main sequence models. With a long orbital period and a small flux ratio (f2/f1) between DI Tau A and B, additional measurements are needed for a better comparison between these observational results and pre-main-sequence models. Finally, we report an average surface magnetic field strength (B¯) for DI Tau A, of ∼0.55 kG, which is unusually low in the context of young active stars. 

Artificial Intelligence (AI) brings advancements to support pathologists in navigating high-resolution tumor images to search for pathology patterns of interest. However, existing AI-assisted tools have not realized the promised potential due to a lack of insight into pathology and HCI considerations for pathologists’ navigation workflows in practice. We first conducted a formative study with six medical professionals in pathology to capture their navigation strategies. By incorporating our observations along with the pathologists’ domain knowledge, we designed NaviPath — a human-AI collaborative navigation system. An evaluation study with 15 medical professionals in pathology indicated that: (i) compared to the manual navigation, participants saw more than twice the number of pathological patterns in unit time with NaviPath, and (ii) participants achieved higher precision and recall against the AI and the manual navigation on average. Further qualitative analysis revealed that participants’ navigation was more consistent with NaviPath, which can improve the examination quality. 

The inductive biases of trained neural networks are difficult to understand and, consequently, to adapt to new settings. We study the inductive biases of linearizations of neural networks, which we show to be surprisingly good summaries of the full network functions. Inspired by this finding, we propose a technique for embedding these inductive biases into Gaussian processes through a kernel designed from the Jacobian of the network. In this setting, domain adaptation takes the form of interpretable posterior inference, with accompanying uncertainty estimation. This inference is analytic and free of local optima issues found in standard techniques such as fine-tuning neural network weights to a new task. We develop significant computational speed-ups based on matrix multiplies, including a novel implementation for scalable Fisher vector products. Our experiments on both image classification and regression demonstrate the promise and convenience of this framework for transfer learning, compared to neural network fine-tuning.