tensorflow is vulnerable to denial of service. Lack of proper specification of a valid SparseTensor
for splits
argument of the functionRaggedBincount
allows an attacker to crash the application and potentially read confidential information from outside the bounds of the splits
tensor buffer.