Dynamic Application Placement Under Service and Memory Constraints

2005 ◽  
pp. 391-402 ◽  
Author(s):  
Tracy Kimbrel ◽  
Malgorzata Steinder ◽  
Maxim Sviridenko ◽  
Asser Tantawi
Keyword(s):  
Application Placement ◽  
Memory Constraints

scholarly journals Instance Sequence Queries for Video Instance Segmentation with Transformers

Sensors ◽  
2021 ◽  
Vol 21 (13) ◽  
pp. 4507
Author(s):  
Zhujun Xu ◽  
Damien Vivet
Keyword(s):  
Data Association ◽  
Video Clip ◽  
Complex Data ◽  
Training Procedure ◽  
Post Processing ◽  
Bipartite Matching ◽  
Frame Method ◽  
Memory Constraints ◽  
Instance Segmentation

Existing methods for video instance segmentation (VIS) mostly rely on two strategies: (1) building a sophisticated post-processing to associate frame level segmentation results and (2) modeling a video clip as a 3D spatial-temporal volume with a limit of resolution and length due to memory constraints. In this work, we propose a frame-to-frame method built upon transformers. We use a set of queries, called instance sequence queries (ISQs), to drive the transformer decoder and produce results at each frame. Each query represents one instance in a video clip. By extending the bipartite matching loss to two frames, our training procedure enables the decoder to adjust the ISQs during inference. The consistency of instances is preserved by the corresponding order between query slots and network outputs. As a result, there is no need for complex data association. On TITAN Xp GPU, our method achieves a competitive 34.4% mAP at 33.5 FPS with ResNet-50 and 35.5% mAP at 26.6 FPS with ResNet-101 on the Youtube-VIS dataset.

Optimal list order under partial memory constraints

1980 ◽  
Vol 17 (4) ◽  
pp. 1004-1015 ◽  
Author(s):  
Y. C. Kan ◽  
S. M. Ross
Keyword(s):  
Decision Maker ◽  
Wide Class ◽  
List Order ◽  
Memory Constraints ◽  
Average Position ◽  
As If

Suppose that we are given a set of n elements which are to be arranged in some order. At each unit of time a request is made to retrieve one of these elements — the ith being requested with probability Pi. We show that the rule which always moves the requested element one closer to the front of the line minimizes the average position of the element requested among a wide class of rules for all probability vectors of the form P1 = p, P2= · ·· = Pn = (1 – p)/(n − 1). We also consider the above problem when the decision-maker is allowed to utilize such rules as ‘only make a change if the same element has been requested k times in a row', and show that as k approaches infinity we can do as well as if we knew the values of the Pi.

Sign in / Sign up

Export Citation Format

Share Document