A Systematic Approach to Design Real-World Human-in-the-Loop Deep Reinforcement Learning: Salient Features, Challenges and Trade-offs
Jalal Arabneydi, Saiful Islam, Srijita Das, Sai Krishna Gottipati, William Duguay, Cloderic Mars, Matthew E. Taylor, Matthew Guzdial, Antoine Fagette, Younes Zerouali · 2025
Read A Systematic Approach to Design Real-World Human-in-the-Loop Deep Reinforcement Learning: Salient Features, Challenges and Trade-offs by Jalal Arabneydi, Saiful Islam, Srijita Das, Sai Krishna Gottipati, William Duguay, Cloderic Mars, Matthew E. Taylor, Matthew Guzdial, Antoine Fagette, Younes Zerouali (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-17006.txt.
Description
Jalal Arabneydi and a large team set out a plan for deep learning rigs that learn with a human in the loop. Trial-and-error AI can steer hard tasks, yet real rigs gain when a human feeds hints mid-run. Their 2025 plan stacks three learning modes: self trial, copying a guide, and carry-over from past tasks. Human input comes in three forms too: score marks, set acts, and shown runs.
A drone duel tests the plan: guard drones must stop foe drones from a barred zone. The rig runs on an award-grade open tool for loop work. Loop runs trained faster and scored higher than lone runs. Hints steered the learning path and cut its shake. Too much or too little hint hurt the end grade, so dose counts. Twin stress tests, a flood and a decoy rush, also passed.
Catalog record
| CREATOR | Jalal Arabneydi, Saiful Islam, Srijita Das, Sai Krishna Gottipati, William Duguay, Cloderic Mars, Matthew E. Taylor, Matthew Guzdial, Antoine Fagette, Younes Zerouali |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-17006