1: \begin{abstract}

2: Reconfigurable intelligent surface (RIS) have been cast as a promising alternative to alleviate blockage vulnerability and enhance coverage capability for terahertz (THz) communications. Owning to large-scale array elements at transceivers and RIS, the codebook based beamforming can be utilized in a computationally efficient manner. However, the codeword selection for analog beamforming is an intractable combinatorial optimization (CO) problem. To this end, by taking the CO problem as a classification problem, a multi-task learning based analog beam selection (MTL-ABS) framework is developed to implement multiple codeword selection tasks concurrently at transceivers and RIS and to accelerate the beam training process. In addition, residual network and self-attention mechanism are used to combat the network degradation and mine intrinsic THz channel features. Finally, the network convergence is analyzed from a blockwise perspective, and numerical results demonstrate that the MTL-ABS framework greatly decreases the beam training overhead and achieves near optimal sum-rate compared with heuristic search based counterparts.

3: \end{abstract}