Scaling Media Machine Studying at Netflix | by Netflix Expertise Weblog | Feb, 2023

Determine 1 – Media Machine Studying Infrastructure

Media Entry: Jasper

Media Characteristic Storage: Amber Characteristic Retailer

Compute Triggering and Orchestration: Amber Compute

Coaching Efficiency

Serving and Looking

Background

Determine 2 – a sequence of body match cuts from Wednesday.

The place we began

Determine 3- The unique Match Slicing pipeline earlier than leveraging media ML infrastructure parts.
SB = 0: [0, 20], 1: [20, 30], 2: [30, 85], …
# the second shot (index 1) was eliminated and so was clip1.mp4
SB_deduped = 0: [0, 20], 2: [30, 85], …
[
# shots with indices 12 and 729 have a high matching score
shot1: 12, shot2: 729, score: 0.96,
# shots with indices 58 and 419 have a low matching score
shot1: 58, shot2: 410, score: 0.02,

]

The issues we confronted

The place we landed

Determine 4 – Match chopping pipeline constructed utilizing media ML infrastructure parts. Interactions between algorithms are expressed as a characteristic mesh, and every Amber Characteristic encapsulates triggering and compute.