1: \begin{abstract}
2: We propose a new variant of Kelley's cutting-plane method
3: %(a Kelley-like method, KLM)
4: for minimizing a nonsmooth convex Lipschitz-continuous function over the Euclidean space.
5: We derive the method through a constructive approach and prove that it attains the optimal rate of convergence
6: for this class of problems.
7: %In addition, we
8: %discuss some other situations %classes of problems
9: %where the approach presented here is applicable.
10: %such as problems equipped with an inexact oracle.
11:
12: %which naturally extends to other classes of problems, such as
13: %problems equipped with an inexact oracle.
14: %An aggregation technique for obtaining a memory-limited version of the method is also presented.
15: \paragraph{Keywords} Nonsmooth convex optimization; Kelley's cutting-plane method; Bundle and subgradient methods; Duality; Complexity; Rate of convergence
16: \end{abstract}