Modern deep learning models require large amounts of accurately annotated data, which is often difficult to satisfy. Hence, weakly supervised tasks, including object localization (WSOL) and detection (WSOD), have recently received attention in the computer vision community. In this paper, we motivate propose foreground (WSFL) task by showing that both WSOL WSOD can be greatly improved if ground...

Given any positive integer $n,$ let $A(n)$ denote the height of $n^{\text{th}}$ cyclotomic polynomial, that is its maximum coefficient in absolute value. It well known unbounded. We conjecture every natural number can arise as value and prove this assuming for pair consecutive primes $p$ $p'$ with $p\ge 127$ we have $p'-p<\sqrt{p}+1.$ also occurs some polynomial show true if Andrica's always $\...

In this note, we aim to present some properties of the space of all weakly fuzzy bounded linear operators, with the Bag and Samanta’s operator norm on Felbin’s-type fuzzy normed spaces. In particular, the completeness of this space is studied. By some counterexamples, it is shown that the inverse mapping theorem and the Banach-Steinhaus’s theorem, are not valid for this fuzzy setting. Also...

This paper describes and compares application of wavelet basis and Block-Pulse functions (BPFs) for solving fractional integro-differential equation (FIDE) with a weakly singular kernel‎. ‎First‎, ‎a collocation method based on Haar wavelets (HW)‎, ‎Legendre wavelet (LW)‎, ‎Chebyshev wavelets (CHW)‎, ‎second kind Chebyshev wavelets (SKCHW)‎, ‎Cos and Sin wavelets (CASW) and BPFs are presented f...