Optimisation of the ATLAS $b$-tagging performance for the 2016 LHC Run

The performance of the current ATLAS $b$-tagging algorithms has been optimised ahead of the 2016 Run-2 data taking. The algorithms used to identify jets containing $b$-hadrons are briefly described and the improvements relative to the 2015 LHC run are highlighted. Particular emphasis is given to the optimisation and performance of the multivariate $b$-tagging algorithm, MV2. Overall for a $b$-tagging efficiency of 77%, the light-quark rejection factor has been held constant while increasing the charm rejection factor by 40%. The performance at high jet $p_T$ has also been improved.