Shannon Wireless’ Post

This paper presents two strategies to #reduce the #complexity of the #alternating #direction method of multipliers when #applied #to #linear #programming (#ADMM-#LP) decoding of #low-#density #parity-#check codes. First, to address the high complexity of computing a projection onto the parity polytope, the complexity bottleneck of ADMM-LP decoding, we propose the #sparse #affine #projection #algorithm (#SAPA). SAPA projects onto the affine hull of χ ≤ d nearby local codewords where the check degree is d and where χ can be significantly smaller than d . Unlike exact projection, SAPA does not require a #water-#filling process, and thus can be implemented with #lower #per-#iteration #complexity. Second, to reduce the number of #effective #iterations needed for ADMM-LP decoding, the authors propose a randomized layered scheduling framework. ----Amirreza Asadzadeh, Anthony Hobley, Frank Kschischang, Stark Draper More details can be found at this link: https://lnkd.in/gnC2fWtX

This paper proposes a #tree #search #algorithm called #successive #cancellation #ordered #search (#SCOS) for #G_N-#coset #codes that implements #maximum-#likelihood (#ML) #decoding with adaptive complexity for transmission over #binary-#input #AWGN #channels. Unlike bit-flip decoders, no outer code is needed to terminate decoding; therefore, SCOS also applies to G_N-coset codes modified with dynamic frozen bits. The average complexity is close to that of #successive #cancellation (#SC) decoding at practical #frame #error #rates (#FERs) for codes with wide ranges of rate and lengths up to 512 bits, which perform within 0.25 dB or less from the random coding union bound and outperform #Reed–Muller codes under ML decoding by up to 0.5 dB.----Peihong Yuan, Mustafa Cemil Coşkun More details can be found at this link: https://lnkd.in/gP_Ae4Mv

In this paper, they consider the problem of recursively designing uniquely decodable ternary code sets for highly overloaded synchronous #code-#division #multiple-#access (#CDMA) systems. The proposed code set achieves larger number of users than any other known state-of-the-art ternary codes that offer low-complexity decoders in the noisy transmission. Moreover, this paper propose a simple decoder that uses only a few comparisons and can allow the user to uniquely recover the information bits. Compared to #maximum #likelihood (#ML) decoder, which has a high computational complexity for even moderate code length, the proposed decoder has much lower computational complexity. They also derived the computational complexity of the proposed recursive decoder analytically. ----Michel Kulhandjian, @Claude D'Amours, Hovannes Kulhandjian More details can be found at this link: https://lnkd.in/gZ9vFyN9

Happy to share that our paper, "Motion-Adaptive Inference for Flexible Learned B-Frame Compression," has been accepted for presentation at IEEE ICIP 2024! We demonstrate state-of-the-art results among other learned codecs, even outperforming the random access mode of H.266 reference codec VTM-18.0. The paper highlights the pivotal role of effective modeling of complex video dynamics and addressing data drift between training and real-world test conditions in the success of our framework. Check it out on arXiv preprint: https://lnkd.in/d4BVZP7H. #IEEE #ICIP2024 #ImageProcessing #VideoProcessing #VideoCoding

The analysis of stochastic electromagnetic fields is gaining more and more relevance due to the exponential growth of complex high-performance electronic systems. Stochastic electromagnetic fields are characterized by auto and cross-correlation functions which can be obtained from experimental data. Different methods have been proposed for the numerical propagation of correlation information within the near-field region of a stochastic radiator. As a guideline for general geometries, near-field Green's functions combined with the method of moments can be used for the numerical estimation of field correlations in the near-field surrounding a device under test. In the ray-tracing limit, a more insightful propagation method based on the Wigner transformation has been devised, through which it is also possible to estimate the propagation of stochastic fields in the near-field. In this paper we report on the implementation of the proposed guide in the open source Python programming language, accessible through the IEEE Standard Association repository to ensure the dissemination of the standard and encourage the development of new versions. Link to paper below:

This paper considers a general framework for #massive #random #access based on #sparse #superposition #coding. The authors provide guidelines for the code design and propose the use of constant-weight codes in combination with a dictionary design based on Gabor frames. The decoder applies an extension of #approximate #message #passing (#AMP) by iteratively exchanging soft information between an AMP module that accounts for the dictionary structure, and a second inference module that utilizes the structure of the involved constant-weight code. They apply the encoding structure to (i) the unsourced random access setting, where all users employ a common dictionary, and (ii) to the “sourced” random access setting with user-specific dictionaries. When applied to a fading scenario, the communication scheme essentially operates non-coherently, as channel state information is required neither at the transmitter nor at the receiver. ---- Patrick Agostini, Zoran Utkovski, Alexis Decurninge, Maxime Guillaud, Slawomir Stanczak More details can be found at this link: https://lnkd.in/eRNu8TuQ

In this episode, we discuss LLM Maybe LongLM: Self-Extend LLM Context Window Without Tuning by Hongye Jin, Xiaotian Han, Jingfeng Yang, Zhimeng Jiang, Zirui Liu, Chia-Yuan Chang, Huiyuan Chen, Xia Hu. The paper presents SelfExtend, a novel method for extending the context window of Large Language Models (LLMs) to better handle long input sequences without the need for fine-tuning. SelfExtend incorporates bi-level attention mechanisms to manage dependencies between both distant and adjacent tokens, allowing LLMs to operate beyond their original training constraints. The method has been tested comprehensively, showing its effectiveness, and the code is shared for public use, addressing the key challenge of LLMs' fixed sequence length limitations during inference.

🚩 #MasterDSA Challenge Update 🚩 Today's achievement: Pascal's Triangle 🔺 Pascal's Triangle is a fundamental concept in combinatorics, where each element is the sum of the two elements directly above it. I implemented a dynamic solution that efficiently generates each row and calculates combinations on the fly. 🔑 Key Insights: 1. Utilized the combinatorics formula to generate each row element. 2. Modularized the solution by creating a helper function to compute a row based on binomial coefficients. 3. Efficient memory usage while ensuring accuracy through precise iteration. 💡 This problem was great for understanding recursion, combinations, and dynamic list building! 📝 Problem: Pascal's Triangle (LeetCode) ⏱️ Runtime: 0 ms (Beats 100% of submissions) 📉 Memory Usage: 8.17 MB (Efficient use of memory) Loving the progress I'm making as part of the #MasterDSA challenge! 🔥 #PascalTriangle #Combinatorics #Mathematics #DataStructures #ProblemSolving #MasterDSA #LeetCode #TechLearning

𝐒𝐏𝐃𝐞𝐬𝐢𝐠𝐧: A method that combines structural sequence profile, fast shape recognition, and pre-trained language models for protein sequence design Paper: https://lnkd.in/gXVurtpE Server & Source Code: https://lnkd.in/gMYwjSgT

Human matting is a foundation task in image and video processing where human foreground pixels are extracted from the input. Prior works either improve the accuracy by additional guidance or improve the temporal consistency of a single instance across frames. This Paper proposes a new framework MaGGIe, Masked Guided Gradual Human Instance Matting, which predicts alpha mattes progressively for each human instance while maintaining the computational cost, precision, and consistency. The method leverages modern architectures, including transformer attention and sparse convolution, to output all instance mattes simultaneously without exploding memory and latency. Code and datasets are available at 👉 https://lnkd.in/g5NGetUY. Paper 👉 https://lnkd.in/gRakdity