Is Casper/Sharding like Delegated Byzantine Fault Tolerance?

From my understanding, Sharding will allow the Ethereum network to dynamically select nodes which will validate transactions concurrently. There will be many clusters working at the same time instead of the current single thread model where every node on the network (currently miners) must validate the transaction.

Does this mean that Sharding is like having multiple Delegated Byzantine Fault Tolerance nodes processing transactions all over Ethereum net? We'll lose some efficiency since the selected nodes will dynamically change but it should scale way more than DBFT networks if this is true. Am I understanding this correctly?

Submitted February 09, 2018 at 11:04PM }
via reddit