scholarly journals The Unreasonable Effectiveness of the Final Batch Normalization Layer

2021 ◽  
pp. 81-93
Author(s):  
Veysel Kocaman ◽  
Ofer M. Shir ◽  
Thomas Bäck
Author(s):  
Sushma Shrestha ◽  
Abeer Alsadoon ◽  
P. W. C. Prasad ◽  
Indra Seher ◽  
Omar Hisham Alsadoon

2021 ◽  
Vol 7 (2) ◽  
pp. 37
Author(s):  
Isah Charles Saidu ◽  
Lehel Csató

We present a sample-efficient image segmentation method using active learning, we call it Active Bayesian UNet, or AB-UNet. This is a convolutional neural network using batch normalization and max-pool dropout. The Bayesian setup is achieved by exploiting the probabilistic extension of the dropout mechanism, leading to the possibility to use the uncertainty inherently present in the system. We set up our experiments on various medical image datasets and highlight that with a smaller annotation effort our AB-UNet leads to stable training and better generalization. Added to this, we can efficiently choose from an unlabelled dataset.


2013 ◽  
Vol 22 (12) ◽  
pp. 1342030 ◽  
Author(s):  
KYRIAKOS PAPADODIMAS ◽  
SUVRAT RAJU

We point out that nonperturbative effects in quantum gravity are sufficient to reconcile the process of black hole evaporation with quantum mechanics. In ordinary processes, these corrections are unimportant because they are suppressed by e-S. However, they gain relevance in information-theoretic considerations because their small size is offset by the corresponding largeness of the Hilbert space. In particular, we show how such corrections can cause the von Neumann entropy of the emitted Hawking quanta to decrease after the Page time, without modifying the thermal nature of each emitted quantum. Second, we show that exponentially suppressed commutators between operators inside and outside the black hole are sufficient to resolve paradoxes associated with the strong subadditivity of entropy without any dramatic modifications of the geometry near the horizon.


2021 ◽  
Author(s):  
Shang-Hua Gao ◽  
Qi Han ◽  
Duo Li ◽  
Ming-Ming Cheng ◽  
Pai Peng
Keyword(s):  

Sign in / Sign up

Export Citation Format

Share Document