Accelerating DNN Architecture Search at Scale Using Selective Weight Transfer

Author(s):  
Hongyuan Liu ◽  
Bogdan Nicolae ◽  
Sheng Di ◽  
Franck Cappello ◽  
Adwait Jog
Keyword(s):  
2001 ◽  
Author(s):  
Badih Jawad ◽  
Hassane El-Khoury ◽  
James Todd ◽  
Mark Masen

2012 ◽  
Vol 27 (2) ◽  
pp. 74-84 ◽  
Author(s):  
Donna Krasnow ◽  
M Virginia Wilmerding ◽  
Shane Stecyk ◽  
Matthew Wyon ◽  
Yiannis Koutedakis

The purpose of this study was to examine grand battement devant at the barre, in the center, and traveling through space. The primary focus was to consider weight transfer in three conditions: from two feet to one foot for the barre and center conditions, and from one foot to the other foot in traveling. Forty female dancers volunteered (mean age 30.0 ± 13.0 yrs) and were placed in three groups: beginner (n = 12), intermediate (n = 14), and advanced (n = 14). Data were collected with a 7-camera Vicon motion capture system using a Plug-in Gait Full Body Marker set and with two Kistler force plates. Dancers executed five grand battement devant in each of three conditions in randomized order: at the barre in 1st position, in the center in 1st position, and traveling through space. Four variables were investigated: center of gravity of the full trunk, center of gravity of the pelvis, center of gravity of the upper trunk, and center of mass. Data were analyzed in three intervals—stance to battement initiation (STN to GBI), initiation to battement peak (GBI to GBP), and peak to end (GBP to END)—and in the x-axis and y-axis. The main effect condition was significant for all four variables in both x-axis and y-axis (p<0.001). There were no significant differences for training and no significant condition 3 training interactions. Condition was significant for all three intervals (STN to GBI, GBI to GBP, and GBP to END) for all four variables in both axes (p<0.01). Dance educators might consider the importance of allocating sufficient time in dance practice to each of the three conditions—barre, center, and traveling—to ensure development of a variety of motor strategies for weight transfer.


2018 ◽  
Vol 8 (16) ◽  
pp. 47-59
Author(s):  
Rasoul Yaali ◽  
Maliheh Naeimi Kia ◽  
Amin Gholami

2016 ◽  
Vol 61 (No. 1) ◽  
pp. 47-53 ◽  
Author(s):  
G.V. Kumar Prasanna

Geometric performance parameters of three-point hitch system of a most sold model of a 2-wheel drive Indian tractor were determined by generating the path of upper and lower hitch points by kinematic linkage analysis. At various locations of pivot point of upper link and adjustments in the length of lift rods, hitch linkage system of the tractor fulfilled all the requirements specified by the standards for category-I and II hitches. An insight into the kinematic linkage analysis revealed that the hitch linkage of the selected tractor is the most suitable for operations with soil working implements. Attachment of upper link to the topmost pivot point reduced the change in orientation of implement during lifting and ensured better weight transfer from implement to the rear axle of tractor. The kinematic linkage analysis has the potential to identify the best settings of the hitch linkage system for the effective utilization of tractor power during various farm operations.


Author(s):  
Jillian Duquesnay ◽  
Jason Baldwin ◽  
James Wesley Rains

An alternative to lifted installation of topsides by a derrick barge is installation of single, integrated offshore platform topsides by floatover method. Floatover installation reduces hook-up and commissioning, which results in overall schedule and cost savings. Numerous papers were written recently to describe many aspects of the floatover operations. Nature of the floatover is such that it requires detailed engineering analyses, numerical simulations, model testing, and planning to evaluate all phases of the operation [Ref 5], [Ref 6]. Proper analysis of floatover requires numerical simulations using time-domain methodology to evaluate the system non-linearities inherent in the floatover hardware, fendering, mooring lines. Normally, weight transfer stages are given a high profile however it is found that the docking and undocking stages are equally as important. These sensitive stages of the floatover operation occur when the barge is entering the jacket slot prior to the floatover and exiting the jacket slot afterwards. The operation is sensitive to the prevailing weather and the number of simulations to make sure the operations can be performed safely is significant. Results of the docking and undocking analyses usually determine the weather standby and thus workability. This paper will address the docking and undocking stages of floatover for a barge that does not have its own propulsion. The paper shall include a concurrent investigation on effects of weather criteria. Stiffness of the hardware, mating lines/cross lines, mooring lines and the effect they have on the system will be discussed.


Sign in / Sign up

Export Citation Format

Share Document