The short answer:

An actor is someone or something that is capable of sending information across your network.

The long answer:
The Ridgeback security appliance is really tracking agents in the sense of philosophy, sociology, linguistics, and artificial intelligence.  That is, an agent is an entity capable of “agency,” or essentially capable of making something happen.  Although I would prefer to talk about agents and agency, the term “agent” did not fly on the marketing front.  The problem is that many products use what they call agents, or software programs installed on endpoints.  When I said “agent,” everyone around me thought, “Oh, those programs we have to install and maintain.”  My using the term “agent” caused quite a bit of confusion.
So, I gave up on describing agents and agency, and started using the term “actor” instead.  I use the term actor in the sense of actants, Interactions of Actors Theory, policy debate, actor-network theory, actor model, or UML.  In a nutshell, an actor is something (including people, processes, and machines) that can do something.
Some concrete examples of actors include an employee using a desktop computer, and automated software patching process, a web server, a web browser, a human using a web browser, and a Javascript computer program running inside of a web browser.  Yes, things get very complicated when one considers all the implications.  Thankfully, all that complexity is what the Ridgeback security appliance handles internally.