The field of quantized linear computations and integrated sensing and communications is rapidly advancing, with a focus on improving the efficiency and accuracy of these systems. Recent research has explored the use of deep learning-based approaches, such as neural operators and attention modules, to enhance the performance of these systems. Additionally, there is a growing interest in the development of hybrid beamfocusing architectures for near-field integrated sensing and communication systems, which can provide improved energy efficiency and hardware efficiency. Noteworthy papers in this area include: Improved Bounds on Access-Redundancy Tradeoffs in Quantized Linear Computations, which presents improved impossibility results for quantized linear computations and initiates the study of approximate recovery for this problem. Channel Estimation for Flexible Intelligent Metasurfaces, which proposes a deep learning-based framework using a Fourier neural operator to learn the continuous operator that maps flexible intelligent metasurface shapes to channel responses. Active IRS-Enabled Integrated Sensing and Communications with Extended Targets, which derives the sensing Cramér-Rao bound for estimating the target response matrix and jointly optimizes the transmit beamforming at the base station and the reflective beamforming at the active intelligent reflecting surface. Deep Joint Source-Channel Coding for Small Satellite Applications, which presents a comprehensive deep joint source-channel coding framework tailored for satellite communications and integrates a realistic, multi-state statistical channel model to guide its training and evaluation. Deep Learning-Based Rate-Adaptive CSI Feedback for Wideband XL-MIMO Systems in the Near-Field Domain, which proposes a rate-adaptive deep learning framework designed to enable efficient channel state information feedback in wideband near-field extremely large-scale MIMO systems.