<p>In cellular networks, an Intelligent Reflecting Surface (IRS) technology has great promise for increasing transmission rates. This improvement happened because a big IRS with a large amount of passive reflecting components was carefully evaluated and calibrated to correctly concentrate the incident beams onto the receiver. Even though, the base station (BS) must effectively collect Channel State Information (CSI) to obtain this beam forming gain. The standard pilot estimation approach is challenging to apply since the count of channel coefficients is directly relative to the number of IRS parts, and the passive intelligent reflecting surface not include Radio Frequency chains. In this paper, an Enhancing Multi-User MIMO Channel Estimation with IRS: Leveraging Data-Aided Methods and Deep Expectation Maximization (IRS-MIMO-DAM-DEM) is proposed. The Data-aided Channel Estimate Model (DAM) for multiple-user MIMO networks is an IRS’s assistance to improve the possible rate by lowering the overhead associated with channel estimation caused by the pilot transmission. In this context, suggest utilizing a traditional pilot transmission method to estimate the direct (User-BS) channels. The Deep Expectation Maximization (DEM) method is used to potentially improve accuracy of channel estimation. The proposed IRS-MIMO-DAM-DEM method attains 10.28%, 18.22%, and 19.27% lower NMSE and 13.76%, 14.82%, and 12.47% lower MMSE compared with existing methods: Semi-Blind Channel Estimation for IRS in Huge MIMO schemes (SBCE-IRS-MIMO), Channel Estimation for Reconfigurable Intelligent Surface-aided Multi-User mm Wave MIMO (IRS-CS-MIMO), and Channel Estimation Overhead Reduction Scheme and impact in IRS-assisted Systems (IRS-JOA-MIMO) respectively.</p>