Well, if these bytes conform to the standard IEEE 754 floating point format, you can easily transfer the bits to float or double. Either...
1. Using the java.lang.Float and java.lang.Double class that has methods to take bits from ints or longs to floats or doubles. So you can use the shifting and AND operators to get the bytes into place and then use those methods.
2. Or you can use the java.nio.ByteBuffer class, which has methods that does what you want -- including taking care of the Big or Little Endian.
Of course, this all assumes that the bytes conform to the floating point standard.