Northwestern University

About: Northwestern University is a education organization based out in Evanston, Illinois, United States. It is known for research contribution in the topics: Population & Medicine. The organization has 75430 authors who have published 188857 publications receiving 9463252 citations. The organization is also known as: Northwestern & NU.

Structural Information from Ion Mobility Measurements: Effects of the Long-Range Potential

Abstract: In a number of recent studies, information about the structure of large polyatomic ions has been deduced from gas phase ion mobility measurements by comparing mobilities measured in helium to those estimated for assumed geometries using a hard sphere projection approximation. To examine the validity of this approach, we have compared mobilities calculated using the hard sphere projection approximation for a range of fullerenes (C20−C240) to those determined from trajectory calculations with a more realistic He−fullerene potential. The He−fullerene potential we have employed, a sum of two-body 6-12 interactions plus a sum of ion-induced dipole interactions, was calibrated using the measured mobility of C60+ in helium over an 80−380 K temperature range. For the systems studied, the long-range interactions between the ion and buffer gas have a small, less than 10%, effect on the calculated mobility at room temperature. However, the effects are not insignificant, and in many cases it will be necessary to cons...

On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima

Abstract: The stochastic gradient descent (SGD) method and its variants are algorithms of choice for many Deep Learning tasks. These methods operate in a small-batch regime wherein a fraction of the training data, say $32$-$512$ data points, is sampled to compute an approximation to the gradient. It has been observed in practice that when using a larger batch there is a degradation in the quality of the model, as measured by its ability to generalize. We investigate the cause for this generalization drop in the large-batch regime and present numerical evidence that supports the view that large-batch methods tend to converge to sharp minimizers of the training and testing functions - and as is well known, sharp minima lead to poorer generalization. In contrast, small-batch methods consistently converge to flat minimizers, and our experiments support a commonly held view that this is due to the inherent noise in the gradient estimation. We discuss several strategies to attempt to help large-batch methods eliminate this generalization gap.

Nivolumab for Metastatic Renal Cell Carcinoma: Results of a Randomized Phase II Trial

Abstract: Purpose Nivolumab is a fully human immunoglobulin G4 programmed death–1 immune checkpoint inhibitor antibody that restores T-cell immune activity. This phase II trial assessed the antitumor activity, dose-response relationship, and safety of nivolumab in patients with metastatic renal cell carcinoma (mRCC). Patients and Methods Patients with clear-cell mRCC previously treated with agents targeting the vascular endothelial growth factor pathway were randomly assigned (blinded ratio of 1:1:1) to nivolumab 0.3, 2, or 10 mg/kg intravenously once every 3 weeks. The primary objective was to evaluate the dose-response relationship as measured by progression-free survival (PFS); secondary end points included objective response rate (ORR), overall survival (OS), and safety. Results A total of 168 patients were randomly assigned to the nivolumab 0.3- (n = 60), 2- (n = 54), and 10-mg/kg (n = 54) cohorts. One hundred eighteen patients (70%) had received more than one prior systemic regimen. Median PFS was 2.7, 4.0, a...