Random Batch Methods for Interacting Particle Systems and Consensus-based Global Non-convex Optimization in High-dimensional Machine Learning (copy)

54 mins 24 secs,  216.24 MB,  WebM  640x360,  29.97 fps,  44100 Hz,  542.72 kbits/sec
Share this media item:
Embed this media item:


About this item
Image inherited from collection
Description: Jin, S
Monday 11th November 2019 - 14:00 to 15:00
 
Created: 2019-11-13 08:43
Collection: Geometry, compatibility and structure preservation in computational differential equations
Publisher: Isaac Newton Institute
Copyright: Jin, S
Language: eng (English)
Distribution: World     (downloadable)
Explicit content: No
Aspect Ratio: 16:9
Screencast: No
Bumper: UCS Default
Trailer: UCS Default
 
Abstract: We develop random batch methods for interacting particle systems with large number of particles. These methods
use small but random batches for particle interactions,
thus the computational cost is reduced from O(N^2) per time step to O(N), for a
system with N particles with binary interactions.
For one of the methods, we give a particle number independent error estimate under some special interactions.
Then, we apply these methods
to some representative problems in mathematics, physics, social and data sciences, including the Dyson Brownian
motion from random matrix theory, Thomson's problem,
distribution of wealth, opinion dynamics and clustering. Numerical results show that
the methods can capture both the transient solutions and the global equilibrium in
these problems.

We also apply this method and improve the consensus-based global optimization algorithm for high
dimensional machine learning problems. This method does not require taking gradient in finding global
minima for non-convex functions in high dimensions.
Available Formats
Format Quality Bitrate Size
MPEG-4 Video 640x360    1.91 Mbits/sec 779.97 MB View Download
WebM * 640x360    542.72 kbits/sec 216.24 MB View Download
iPod Video 480x270    491.44 kbits/sec 195.75 MB View Download
MP3 44100 Hz 249.79 kbits/sec 99.59 MB Listen Download
Auto (Allows browser to choose a format it supports)