There are several hospitals H$1,...,$Hk$.$ Each hospital Hi holds a set of patient

records Di$,1<=$i$<=$k$.$ Note that these training data sets are stored at different locations. Since each data

set Di contains a small number of records, it does not provide sufficient information for learning a strong

classifier. While combining all the data contained in D$1,...,$Dk can help learn a stronger classifier,

unfortunately, due to privacy policies, it is not possible to collect the union D$1+...+$Dk in one place to

learn from the combined data. Using the concepts discussed in class, briefly describe a learning method

that can benefit from the rich information contained in all of D$1,...,$Dk without collecting these data in

one place. Explain how your method achieves these goals.