![](/sites/default/files/styles/workshop_banner_sm_1x/public/crypto_logo.png.jpg?itok=FSK_UXMC)
Abstract
We show how the Blum-Kalai-Wasserman algorithm for Learning With Errors can be modified to require sub-exponential time when the secret is small. This gives the first subexponential algorithm for several other problems, such as subset-sum, NTRU and new lattice problems.