Abstract: We introduce a novel gradient-based approach for solving sequential tasks by dynamically adjusting the underlying myopic potential field in response to feedback and the world's regularities.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results