Presentor: Shy-El Cohen Name: Tal Shuster Name: Alon Helvits

Presentor: Shy-El Cohen Supervisor: Dr. Eliya Nachmani Degree: MSc in computers and electrical engineering Seminar Subject: Hybrid Mamba–Transformer Decoder for Error-Correcting Codes Abstract: We introduce a novel deep learning method for decoding error correction codes based on the Mamba architecture, enhanced with Transformer layers. Our approach proposes a hybrid decoder that leverages Mamba’s efficient sequential modeling while maintaining the global context capabilities of Transformers. To further improve performance, we design a novel layer-wise masking strategy applied to each Mamba layer, allowing selective attention to relevant code features at different depths. Additionally, we introduce a progressive layer-wise loss, supervising the network at intermediate stages and promoting robust feature extraction throughout the decoding process. Comprehensive experiments across a range of linear codes demonstrate that our method outperforms or matches Transformer-only decoders while improving complexity. Name: Tal Shuster Supervisor: Dr. Eliya Nachmani Degree: MSc in computers and electrical engineering Subject of the seminar: Q2D2: A geometry-aware audio codec leveraging two-dimensional quantization Abstract: Recent neural audio codecs have achieved impressive reconstruction quality, typically relying on quantization methods such as Residual Vector Quantization (RVQ), Vector Quantization (VQ) and Finite Scalar Quantization (FSQ). However, these quantization techniques limit the geometric structure of the latent space, make it harder to capture correlations between features leading to inefficiency in representation learning, codebook utilization and token rate. In this paper we introduce Two-Dimensional Quantization (Q2D2), a quantization scheme in which feature pairs are projected onto structured 2D grids, such as hexagonal, rhombic, or rectangular tiling and quantized to the nearest grid values, yielding an implicit codebook defined by the product of grid levels, with codebook sizes comparable to conventional methods. Despite its simple geometric formulation, Q2D2 improves audio compression efficiency, with low token rates and high codebook utilization while maintaining state of the art reconstruction quality. Specifically, Q2D2 achieves competitive to superior performance in various objective and subjective reconstruction metrics, across extensive experiments in speech domain compared to state of the art models. Comprehensive ablation studies further confirm the effectiveness of our design choices. Name: Alon Helvits Degree: Electrical and Computer Engineering Supervisor: Eliya Nachmani Seminar Subject: Score Based Error Correcting Code Decoder Error-correcting codes are fundamental to reliable communication, but practical soft decoding remains difficult to scale across diverse code families and block lengths. In this seminar, I will present a score-based decoding approach that reframes decoding as a continuous-time denoising problem in symbol space (BPSK over AWGN). Instead of relying on magnitude-only reliability features or requiring explicit SNR estimation, the method learns a denoising field directly from the signed channel observation and uses it and parity constraint (syndrome) to define a probability-flow ODE. Decoding proceeds by numerically integrating this ODE from the noisy received vector toward a parity-consistent codeword, with the solver budget providing a natural accuracy–latenc