1: \begin{abstract}

2: We consider the  difference-of-convex (DC) programming problems whose objective function is level-bounded. The classical DC algorithm (DCA) is well-known for solving this kind of problems, which returns a critical point. Recently, de Oliveira and Tcheo incorporated the inertial-force procedure into DCA (InDCA) for potential acceleration and preventing the algorithm from converging to a critical point which is not $d$(directional)-stationary. In this paper, based on InDCA, we propose two refined inertial DCA (RInDCA) with enlarged inertial step-sizes for better acceleration. We demonstrate the subsequential convergence of our refined versions to a critical point. In addition, by assuming the Kurdyka-Łojasiewicz (KL) property of the objective function, we establish the sequential convergence of RInDCA. Numerical simulations on image restoration problem show the benefit of enlarged step-size. 

3: \keywords{Difference-of-convex programming \and Refined Inertial DCA \and Enlarged inertial step-size \and Kurdyka-Łojasiewicz property \and Image restoration}

4: %\and projection operator \and proximity operator

5: % \PACS{PACS code1 \and PACS code2 \and more}

6: % \subclass{MSC code1 \and MSC code2 \and more}

7: \subclass{90C26 \and 90C30 \and 68U10}

8: \end{abstract}

9: