Simone Cazzaniga, Dagmar Simon · British Journal of Dermatology · 2023
Journal Article Accepted manuscript Decoding Atopic Dermatitis: Unveiling Phenotypes Through Data-Driven Methods Get access Simone Cazzaniga, Simone Cazzaniga Department of Dermatology, Inselspital University Hospital of Bern, Bern, SwitzerlandCentro Studi GISED, Bergamo, Italy Correspondence: Simone Cazzaniga, Email:simone.cazzaniga@gised.it https://orcid.org/0000-0001-8161-6138 Search for other works by this author on: Oxford Academic Google Scholar Dagmar Simon Dagmar Simon Department of Dermatology, Inselspital University Hospital of Bern, Bern, Switzerland https://orcid.org/0000-0001-8965-9407 Search for other works by this author on: Oxford Academic Google Scholar British Journal of Dermatology, ljad453, https://doi.org/10.1093/bjd/ljad453 Published: 17 November 2023 Article history Editorial decision: 10 November 2023 Received: 10 November 2023 Accepted: 16 November 2023 Published: 17 November 2023
Cited by 1
Dimitrios Triantafyllou · 2013
In this paper, Numerical Linear Algebra techniques for data encoding and decoding are presented. The algorithms are based on matrix triangularization leading to an ecient coding procedure. Rounding o errors can cause serious problems in the process of coding and thus the decoding may be led to inaccurate results. Techniques for improving the stability of the whole process are proposed. The complexity and the error analysis of the proposed methods are discussed. Examples illustrating the algorithms are presented.
Cited by 4
Association for Computational Linguistics 2025, Matan Avitan, Moran Baruch et al. · Underline Science Inc. · 2025
Large language models (LLMs) power modern AI applications, but processing sensitive data on untrusted servers raises privacy concerns. Homomorphic encryption (HE) enables computation on encrypted data for secure inference. However, neural text generation requires decoding methods like argmax and sampling, which are non-polynomial and thus computationally expensive under encryption, creating a significant performance bottleneck. We introduce cutmax, an HE-friendly argmax algorithm that reduces ciphertext operations compared to prior methods, enabling practical greedy decoding under encryption. We also propose the first HE-compatible nucleus (top-p) sampling method, leveraging cutmax for efficient stochastic decoding with provable privacy guarantees. Both techniques are polynomial, supporting efficient inference in privacy-preserving settings. Moreover, their differentiability facilitates gradient-based sequence-level optimization as a polynomial alternative to straight-through estimators. We further provide strong theoretical guarantees for cutmax, proving its convergence via exponential amplification of the gap ratio between the maximum and runner-up elements. Evaluations on realistic LLM outputs show latency reductions of $24\times$–$35\times$ over baselines, advancing secure text generation.
Cited by 0Open access
Foodeei, Darius, Fan, Simin, Jaggi, Martin · ArXiv.org · 2025
This study investigates semantic uncertainty in large language model (LLM) outputs across different decoding methods, focusing on emerging techniques like speculative sampling and chain-of-thought (CoT) decoding. Through experiments on question answering, summarization, and code generation tasks, we analyze how different decoding strategies affect both the diversity and reliability of model outputs. Our findings reveal that while CoT decoding demonstrates higher semantic diversity, it maintains lower predictive entropy, suggesting that structured exploration can lead to more confident and accurate outputs. This is evidenced by a 48.8% improvement in code generation Pass@2 rates, despite lower alignment with reference solutions. For summarization tasks, speculative sampling proved particularly effective, achieving superior ROUGE scores while maintaining moderate semantic diversity. Our results challenge conventional assumptions about trade-offs between diversity and accuracy in language model outputs, demonstrating that properly structured decoding methods can increase semantic exploration while maintaining or improving output quality. These findings have significant implications for deploying language models in practical applications where both reliability and diverse solution generation are crucial.
Cited by 0Open access
Chun-Ling Kei · 2002
Efficient complexity reduction methods for short-frame iterative decoding by Kei Chun-Ling thesis 2002 xiii, 91 leaves : ill. ; 30 cm Turbo coding has already become a prominent error correction technique for wireless communications, especially after its adaptation in the…Read more ›
Cited by 0
Ni̇hal Taş, Sümeyra Uçar, Nihal Yılmaz Özgür · arXiv (Cornell University) · 2017
We obtain a new coding and decoding method using the generalized Pell $(p,i)$ -numbers. The relations among the code matrix elements, error detection and correction have been established for this coding theory. We give two new blocking algorithms using Pell numbers and generalized Pell $(p,i)$-numbers.
Cited by 0Open access
Rayuwati, Mahmud Mustapa, Askar Taliang et al. · Journal of Physics Conference Series · 2019
Abstract The main problem to be an obstacle in communication between platforms, networks such as local networks or the Internet is a problem in data transfer. The problem often found in a hardware which is cannot receive data that is processed by other hardware. So to change data between two computer systems, it is necessary to convert data from one form to another. One solution to this problem is transfered data or file, previously converted to the plaintext form first. Conversion to the plaintext form can be supported by most computer systems. In this study, a software was designed using Visual Basic programming language that can convert binary files to plaintext using the Base 64 encoding or decoding method, UUEncode / UUDecode, XXEncode / XXDecode, BinHex 4.0, Hexadecimal, and Quoted-Printable. The results of the design is produce a software that can be used as an alternative program to convert files into text files, besides that the program can also display a benchmark comparison graph (speed) of each method and comparison of overhead (the ratio of the size of the file to the file before encoding).
Cited by 2Open access
Matan Avitan, Moran Baruch, Nir Drucker et al. · ArXiv.org · 2025
Large language models (LLMs) power modern AI applications, but processing sensitive data on untrusted servers raises privacy concerns. Homomorphic encryption (HE) enables computation on encrypted data for secure inference. However, neural text generation requires decoding methods like argmax and sampling, which are non-polynomial and thus computationally expensive under encryption, creating a significant performance bottleneck. We introduce cutmax, an HE-friendly argmax algorithm that reduces ciphertext operations compared to prior methods, enabling practical greedy decoding under encryption. We also propose the first HE-compatible nucleus (top-p) sampling method, leveraging cutmax for efficient stochastic decoding with provable privacy guarantees. Both techniques are polynomial, supporting efficient inference in privacy-preserving settings. Moreover, their differentiability facilitates gradient-based sequence-level optimization as a polynomial alternative to straight-through estimators. We further provide strong theoretical guarantees for cutmax, proving its convergence via exponential amplification of the gap ratio between the maximum and runner-up elements. Evaluations on realistic LLM outputs show latency reductions of 24x-35x over baselines, advancing secure text generation.
Cited by 0Open access
BERTHET, Antoine Olivier, ALQUIE, Georges · OpenGrey (Institut de l'Information Scientifique et Technique) · 2001
Cited by 0Open access
Chandanamali Punchihewa · The Sri Lanka Journal of Haematology · 2024
Genetics play a significant role in many haematological disorders, both malignant and otherwise. Many such disorders have been well characterised at the genomic level as a result of research-driven discoveries of disease-causing variants. In addition to disease diagnostics, modern-day clinical molecular genetics span from predictive genomic testing to personalised treatment strategies and post-treatment monitoring. Coupled with the growth of applications are the development in genetic testing technologies, with ever-dropping sequencing costs that has brought gene panels and whole exome or genome sequencing to the standard clinical practice. However, despite the rapid advances in technology, there is still an important place in the clinic, even for the most basic PCR tests from the distant past. With the inherent advantages and limitations of each technology, applying the correct method to the correct test is important not only in the clinical and scientific perspective but also in the economies of healthcare. This article outlines the various genetic techniques and their current and potential future applications in haematology.
Cited by 0Open access
Hammond, Lauren Renee · eScholarship (California Digital Library) · 2023
Radical Modes and Methods: Decoding and Dis-Imprinting Trauma explores women-of-color’s (WOC) engagement with new, hypertextual media to consider the influence these interactions have on literary expression and form as well as self-making. I analyze Do-It-Yourself (DIY) practice(s) and politics of (re)articulation to uncover how counter-cultural engagements and (re)productions enact social transformations via coalitional feminist impulses. I (re)position the hypertext as what I call a “multilithic tool”– a tool of selfhood with which to create alternative, equitable, inclusive, and prosperous futures– given it is a complex process of producing culture in terms of multiple, intersecting or disintersecting, complementary and contradictory, instances of imprinting. I work with and within coalitional feminist practices to center WOC’s hypertextual histories with specific emphasis on Black feminist technocultural production: including hashtag activisms (e.g., #BLM, #MeToo, #SayHerName), punk ezine culture (i.e., Shotgun Seamstress), and Carol Parrot Blue’s The Dawn at My Back. I explore how the WOC who are engaging with these projects are formatting multiple ways of moving through traumatic pain/memories by forging modalities of print and (re)printing a new subjectivity for themselves. Throughout these hypertext projects, I argue, trauma is remediated as a means of “dis-imprinting trauma” altogether via the hypertext’s plasticity.\n
Cited by 0Open access
Foodeei, Darius, Fan, Simin, Jaggi, Martin · arXiv (Cornell University) · 2025
This study investigates semantic uncertainty in large language model (LLM) outputs across different decoding methods, focusing on emerging techniques like speculative sampling and chain-of-thought (CoT) decoding. Through experiments on question answering, summarization, and code generation tasks, we analyze how different decoding strategies affect both the diversity and reliability of model outputs. Our findings reveal that while CoT decoding demonstrates higher semantic diversity, it maintains lower predictive entropy, suggesting that structured exploration can lead to more confident and accurate outputs. This is evidenced by a 48.8% improvement in code generation Pass@2 rates, despite lower alignment with reference solutions. For summarization tasks, speculative sampling proved particularly effective, achieving superior ROUGE scores while maintaining moderate semantic diversity. Our results challenge conventional assumptions about trade-offs between diversity and accuracy in language model outputs, demonstrating that properly structured decoding methods can increase semantic exploration while maintaining or improving output quality. These findings have significant implications for deploying language models in practical applications where both reliability and diverse solution generation are crucial.
Cited by 0Open access
H.-D. Lin, David G. Messerschmitt · IEEE Transactions on Communications · 1993
The parallel Viterbi decoding is discussed for two different cases: uncontrollable sources and controllable sources. For general, uncontrollable Markov processes, a previously known parallel method is extended to a hierarchical parallel decoding approach, which achieves a lower latency. For controllable Markov sources in telecommunications applications, new parallel decoding methods are obtained by controlling the source processes in appropriate ways. The focus is on the parallel decoding methods for controllable sources because these methods have zero processing overhead. Because the methods modify the coding process, they bring positive changes to framing and negative changes to latency and code performance. However, one can adjust the parameters of the methods to make the degradation negligible. Because of their low overhead, the methods are most attractive for high-speed decoders for convolutional and trellis codes, and they are also applicable to other sequential algorithms for suboptimal decoding and estimation of complex Markov sources.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
Cited by 8
Schomburg, Dirk, Bernarding, Johannes · 2020
Cited by 0Open access
Grootswagers, Tijl · Figshare · 2019
Brain decoding methods have transformed the field of Cognitive Neuroscience in the last two decades. As this field has matured, researchers are now asking challenging questions about the interpretation of decoding methods, such as whether the information decoded from neuroimaging data is used by the brain for behaviour. The focus of this thesis is to address the challenge of linking information measured with decoding methods to human behaviour. In the first empirical chapter, using magneto-encephalography (MEG), I describe a broad set of options for conducting time series decoding studies and test the effects that different options in the decoding analysis pipeline can have on the experimental results. The results show that decisions made at all stages of the analysis can significantly affect the results and interpretation of decoding studies. In the second empirical chapter, I explore the distance to bound model as a method for linking brain decoding with behaviour. Using MEG decoding, I tested whether this model can account for behavioural changes in reaction time for categorising degraded objects by animacy. I found that the distance to bound model successfully predicted reaction time, accuracy, and decision time parameters derived from a prominent model of decision making. These findings provide evidence for a systematic relationship between decoded brain representations and perceptual decision-making behaviour. In the third empirical chapter, I examine the distinction between decodable information, and information that can be used in behaviour. Using a searchlight approach on functional Magnetic Resonance Imaging (fMRI) data, I first investigated where decodable information existed in the brain. Secondly, I assessed where in the brain the decoded information was suitable for "read out" by the brain for behaviour. I found that behaviour can only be predicted from a subset of the locations that had decodable information. These results highlight the distinction between decodable information, and information that is relevant for behaviour in the brain. In conclusion, this thesis advances current knowledge on brain decoding methods and on approaches to relating brain representations to behaviour, which is a fundamental challenge in cognitive neuroscience. The results show that decodable information has to be interpreted with caution, and emphasize that continuing to develop methods for linking neuroimaging to behaviour is critical for advancing our understanding of the brain.
Cited by 0Open access
Grootswagers, Tijl · Macquarie University · 2019
Brain decoding methods have transformed the field of Cognitive Neuroscience in the last two decades. As this field has matured, researchers are now asking challenging questions about the interpretation of decoding methods, such as whether the information decoded from neuroimaging data is used by the brain for behaviour. The focus of this thesis is to address the challenge of linking information measured with decoding methods to human behaviour. In the first empirical chapter, using magneto-encephalography (MEG), I describe a broad set of options for conducting time series decoding studies and test the effects that different options in the decoding analysis pipeline can have on the experimental results. The results show that decisions made at all stages of the analysis can significantly affect the results and interpretation of decoding studies. In the second empirical chapter, I explore the distance to bound model as a method for linking brain decoding with behaviour. Using MEG decoding, I tested whether this model can account for behavioural changes in reaction time for categorising degraded objects by animacy. I found that the distance to bound model successfully predicted reaction time, accuracy, and decision time parameters derived from a prominent model of decision making. These findings provide evidence for a systematic relationship between decoded brain representations and perceptual decision-making behaviour. In the third empirical chapter, I examine the distinction between decodable information, and information that can be used in behaviour. Using a searchlight approach on functional Magnetic Resonance Imaging (fMRI) data, I first investigated where decodable information existed in the brain. Secondly, I assessed where in the brain the decoded information was suitable for "read out" by the brain for behaviour. I found that behaviour can only be predicted from a subset of the locations that had decodable information. These results highlight the distinction between decodable information, and information that is relevant for behaviour in the brain. In conclusion, this thesis advances current knowledge on brain decoding methods and on approaches to relating brain representations to behaviour, which is a fundamental challenge in cognitive neuroscience. The results show that decodable information has to be interpreted with caution, and emphasize that continuing to develop methods for linking neuroimaging to behaviour is critical for advancing our understanding of the brain.
Cited by 0Open access
M.P.C. Fossorier, Shu Lin, Dojun Rhee · IEEE Transactions on Information Theory · 1998
In this correspondence, the bit-error probability P/sub b/ for maximum-likelihood decoding of binary linear block codes is investigated. The contribution P/sub b/(j) of each information bit j to P/sub b/ is considered and an upper bound on P/sub b/(j) is derived. For randomly generated codes, it is shown that the conventional approximation at high SNR P/sub b//spl ap/(d/sub H//N).P/sub s/, where P/sub s/ represents the block error probability, holds for systematic encoding only. Also systematic encoding provides the minimum P/sub b/ when the inverse mapping corresponding to the generator matrix of the code is used to retrieve the information sequence. The bit-error performances corresponding to other generator matrix forms are also evaluated. Although derived for codes with a generator matrix randomly generated, these results are shown to provide good approximations for codes used in practice. Finally, for soft-decision decoding methods which require a generator matrix with a particular structure such as trellis decoding, multistage decoding, or algebraic-based soft-decision decoding, equivalent schemes that reduce the bit-error probability are discussed. Although the gains achieved at practical bit-error rates are only a fraction of a decibel, they remain meaningful as they are of the same orders as the error performance differences between optimum and suboptimum decodings. Most importantly, these gains are free as they are achieved with no or little additional circuitry which is transparent to the conventional implementation.
Cited by 44
Chaudhari Sachin, Patel Hardik, E Arikan et al. · International Journal of Science and Research (IJSR) · 2016
The first provably capacity achieving codes with low complexity named polar codes were discovered recently and the successive cancellation (SC) decoding is widely known decoding algorithm for polar codes. There are many techniques like folded SC and permuted SC available for improve the SC decoding algorithm. In this paper, we study the different methods of SC decoding for polar codes which are recently developed and give a comparison based on bit error rate (BER) and decoding complexity.
Cited by 1Open access
MEI Hongyan GENG Yaogang · SHILAP Revista de lepidopterología · 2022
In recent years, image caption generation, as a multimodal task in the field of artificial intelligence, integrates the related research of computer vision and natural language processing, and can realize the modal conversion from image to text. It plays an important role in visual assistance and image understanding, and has attracted extensive attention from researchers. Firstly, this paper describes the task of image caption generation, and introduces three image caption generation methods: template-based method, retrieval-based method and encode-decode method. Their respective method ideas, representative research and advantages and disadvantages are also introduced. Secondly, from the model structure, the research progress of image understanding phase and caption generation phase, this paper expounds in detail the method based on encoding-decoding, and summarizes the research over years into the research of image understanding and caption generation. Image understanding research includes attention mechanism and semantic aspects. The research of caption generation is divided into traditional caption generation, dense caption generation and stylish caption generation. The performance, advantages and disadvantages of the model are summarized, and the datasets and evaluation index of the performance evaluation of the image captioning model are introduced. Finally, the challenges and difficulties in the field of image captioning are pointed out.
Cited by 0Open access
Shaz Furniturewala, Kokil Jaidka, Yashvardhan Sharma · arXiv (Cornell University) · 2024
To be included into chatbot systems, Large language models (LLMs) must be aligned with human conversational conventions. However, being trained mainly on web-scraped data gives existing LLMs a voice closer to informational text than actual human speech. In this paper, we examine the effect of decoding methods on the alignment between LLM-generated and human conversations, including Beam Search, Top K Sampling, and Nucleus Sampling. We present new measures of alignment in substance, style, and psychometric orientation, and experiment with two conversation datasets. Our results provide subtle insights: better alignment is attributed to fewer beams in Beam Search and lower values of P in Nucleus Sampling. We also find that task-oriented and open-ended datasets perform differently in terms of alignment, indicating the significance of taking into account the context of the interaction.
Cited by 0Open access