Bergman metrics as pull-backs of the Fubini-Study metric
February 26, 2023
|
Mathematics
Complex Variables
Domains and more generally complex manifolds whose Bergman metrics have
constant holomorphic sectional curvature are characterized. Our approach is to
treat the Bergman metrics as the pull-back by the Bergman-Bochner maps of the
Fubini-Stud...
Performance is not enough: the story told by a Rashomon quartet
February 26, 2023
| | |
Statistics
Computer Science
Machine Learning
Machine Learning
Applications
The usual goal of supervised learning is to find the best model, the one that
optimizes a particular performance measure. However, what if the explanation
provided by this model is completely different from another model and different
again...
Training neural networks with structured noise improves classification
and generalization
February 26, 2023
|
Physics
Computer Science
Disordered Systems and Neural Networks
Machine Learning
The beneficial role of noise-injection in learning is a consolidated concept
in the field of artificial neural networks, suggesting that even biological
systems might take advantage of similar mechanisms to optimize their
performance. The t...
CrystalBox: Future-Based Explanations for Input-Driven Deep RL Systems
February 26, 2023
| |
Computer Science
Electrical Engineering and Systems Science
Machine Learning
Networking and Internet Architecture
Systems and Control
Systems and Control
We present CrystalBox, a novel, model-agnostic, posthoc explainability
framework for Deep Reinforcement Learning (DRL) controllers in the large family
of input-driven environments which includes computer systems. We combine the
natural deco...
Quasi-F-splittings in birational geometry II
February 26, 2023
| | | | |
Mathematics
Algebraic Geometry
Commutative Algebra
Over an algebraically closed field of characteristic p>41, we prove that
three-dimensional Q-factorial affine klt varieties are
quasi-F-split. Furthermore, we show that the bound on the characteristic is
optimal....
A low latency attention module for streaming self-supervised speech
representation learning
February 26, 2023
| | | |
Computer Science
Electrical Engineering and Systems Science
Sound
Computation and Language
Machine Learning
Audio and Speech Processing
The transformer is a fundamental building block in deep learning, and the
attention mechanism is the transformer's core component. Self-supervised speech
representation learning (SSRL) represents a popular use-case for the
transformer archi...
Arrow-Debreu Meets Kyle: Price Discovery for Derivatives
February 26, 2023
|
Quantitative Finance
Economics
Economics
General Economics
We analyze price discovery in a model where an agent has general private
information regarding state probabilities and trades state-contingent claims.
Our model unifies the elements of Arrow and Debreu (1954) and Kyle (1985). In
the equival...
A Survey on Uncertainty Quantification for Deep Learning: An Uncertainty
Source Perspective
February 26, 2023
|
Computer Science
Statistics
Machine Learning
Machine Learning
Deep neural networks (DNNs) have achieved tremendous success in making
accurate predictions for computer vision, natural language processing, as well
as science and engineering domains. However, it is also well-recognized that
DNNs sometime...
Poissonian pair correlation for directions in multi-dimensional affine
lattices, and escape of mass estimates for embedded horospheres
February 26, 2023
|
Mathematics
Number Theory
Dynamical Systems
We prove the convergence of moments of the number of directions of affine
lattice vectors that fall into a small disc, under natural Diophantine
conditions on the shift. Furthermore, we show that the pair correlation
function is Poissonian ...
PDFO: A Cross-Platform Package for Powell's Derivative-Free Optimization
Solvers
February 26, 2023
|
Mathematics
Optimization and Control
The late Professor M. J. D. Powell devised five trust-region methods for
derivative-free optimization, namely COBYLA, UOBYQA, NEWUOA, BOBYQA, and
LINCOA. He carefully implemented them into publicly available solvers, renowned
for their robu...