Mehdi Adibi (505445), James S. McDonald (340978), Colin W. G. Clifford (347632) et al. · Figshare · 2015
<p><b>A.</b> The value of <i>A</i> for the optimal decoder when optimized on population responses under 6 µm adaptation, against the value of <i>A</i> for the optimal decoder when optimized on non-adapted population responses. <b>B.</b> As in <b>A</b>, but for 12 µm adaptation. <b>C.</b> The per cent improvement in the value of <i>A</i> under 6 µm adaptation relative to the non-adapted state. <b>D.</b> As in <b>C</b>, but for 12 µm adaptation. <b>E.</b> The effect of 6 µm adaptation on the value of <i>A</i> for every stimulus pair, in a sample session with 11 simultaneously recorded single neurons, indicated by purple color in <b>A</b>–<b>D</b>. The upper triangle corresponds to the groupwise-optimal decoder, while the lower triangle corresponds to the pairwise-optimal decoder. <b>F.</b> As in <b>E</b>, but for 12 µm adaptation.</p>
Cited by 0Open access
Xianren Wu, Hamid R. Sadjadpour (contact, Zhi Tian · 2004
In this paper, we propose a new two-stage structure for computationally efficient Maximum-Likelihood decoding (MLD) of linear block codes. With this structure, near optimal MLD performance can be achieved at low complexity through two-stage processing.The first stage of processing estimates a minimum sufficient set (MSS) of candidate codewords that contains the optimal codeword, while the second stage performs optimal or sub-optimal decoding search within the estimated MSS of small size. Based on the new structure, we propose a decoding algorithm that systematically trades off between the decoding complexity and the bounded block error rate performance. A low-complexity complementary decoding algorithm is developed to estimate the MSS, followed by an ordered algebraic decoding (OAD) algorithm to achieve flexible system design. Since the size of the MSS changes with the signal-to-noise ratio (SNR), the overall decoding complexity adaptively scales with the quality of the communication link. Theoretical analysis is provided to evaluate the potential complexity reduction enabled by the proposed decoding structure. I.
Cited by 0
Chloe Su, Zhe Ye, Samuel Tenka et al. · ArXiv.org · 2026
Decoding from large language models (LLMs) typically relies on fixed sampling hyperparameters (e.g., temperature, top-p), despite substantial variation in task difficulty and uncertainty across prompts and individual decoding steps. We propose to learn adaptive decoding policies that dynamically select sampling strategies at inference time, conditioned on available compute resources. Rather than fine-tuning the language model itself, we introduce lightweight decoding adapters trained with reinforcement learning and verifiable terminal rewards (e.g. correctness on math and coding tasks). At the sequence level, we frame decoding as a contextual bandit problem: a policy selects a decoding strategy (e.g. greedy, top-k, min-p) for each prompt, conditioned on the prompt embedding and a parallel sampling budget. At the token level, we model decoding as a partially observable Markov decision process (POMDP), where a policy selects sampling actions at each token step based on internal model features and the remaining token budget. Experiments on the MATH and CodeContests benchmarks show that the learned adapters improve the accuracy-budget tradeoff: on MATH, the token-level adapter improves Pass@1 accuracy by up to 10.2% over the best static baseline under a fixed token budget, while the sequence-level adapter yields 2-3% gains under fixed parallel sampling. Ablation analyses support the contribution of both sequence- and token-level adaptation.
Cited by 0Open access
Chloe Su, Zhe Ye, Samuel Tenka et al. · arXiv (Cornell University) · 2026
Decoding from large language models (LLMs) typically relies on fixed sampling hyperparameters (e.g., temperature, top-p), despite substantial variation in task difficulty and uncertainty across prompts and individual decoding steps. We propose to learn adaptive decoding policies that dynamically select sampling strategies at inference time, conditioned on available compute resources. Rather than fine-tuning the language model itself, we introduce lightweight decoding adapters trained with reinforcement learning and verifiable terminal rewards (e.g. correctness on math and coding tasks). At the sequence level, we frame decoding as a contextual bandit problem: a policy selects a decoding strategy (e.g. greedy, top-k, min-p) for each prompt, conditioned on the prompt embedding and a parallel sampling budget. At the token level, we model decoding as a partially observable Markov decision process (POMDP), where a policy selects sampling actions at each token step based on internal model features and the remaining token budget. Experiments on the MATH and CodeContests benchmarks show that the learned adapters improve the accuracy-budget tradeoff: on MATH, the token-level adapter improves Pass@1 accuracy by up to 10.2% over the best static baseline under a fixed token budget, while the sequence-level adapter yields 2-3% gains under fixed parallel sampling. Ablation analyses support the contribution of both sequence- and token-level adaptation.
Cited by 0Open access
Mohammad H. Taghavi, et al. · 2006
The ability of linear programming (LP) decoding to detect failures, and its potential for improvement by the addition of new constraints, motivates the use of an adaptive approach in selecting the constraints for the underlying LP problem. In this paper, we show that the application of such adaptive methods can significantly reduce the complexity of the LP decoding algorithm, which, in the standard formulation, is exponential in the maximum row weight of the parity-check matrix. We further show that adaptively adding new constraints, e.g. by combining parity checks, can provide large gains in LP decoder performance.
Cited by 0
Dr. Kavita Nalini*, Rohan Kumar · Zenodo (CERN European Organization for Nuclear Research) · 2019
Conventional encoder is widely applied to lots of wireless communications including 3G/4G communication and digital video Broad casting (DVB), IOT (Internet of things) transmission and so on. A reconfigurable Viterbi decoder design is proposed for LTE-A, WiMAX, LTECDMA, GSM and TD-SCDMA. The proposed flexible architecture supports a polynomial reconfiguration. More over both tail biting and zero trellis terminating modes are supported. The transmitter encoder and receiver decoder are designed for verifying functionality .The purpose is to determine the feasible to design data error correcting to reduce the decoding latency and complexity it employs forward trace back method and sliding window technology.
Cited by 0Open access
Dr. Kavita Nalini*, Rohan Kumar · Zenodo (CERN European Organization for Nuclear Research) · 2019
Conventional encoder is widely applied to lots of wireless communications including 3G/4G communication and digital video Broad casting (DVB), IOT (Internet of things) transmission and so on. A reconfigurable Viterbi decoder design is proposed for LTE-A, WiMAX, LTECDMA, GSM and TD-SCDMA. The proposed flexible architecture supports a polynomial reconfiguration. More over both tail biting and zero trellis terminating modes are supported. The transmitter encoder and receiver decoder are designed for verifying functionality .The purpose is to determine the feasible to design data error correcting to reduce the decoding latency and complexity it employs forward trace back method and sliding window technology.
Cited by 0Open access
Jianping Li, Cai Chaoshi, Yunfei Feng · 2008
The extrinsic information and the received systematic observations exert different influence on decision-making with the change of SNR values for turbo iterative decoding. By multipling a weighted coefficient for the priori information, this paper proposed a novel adaptive iterative soft decoding algorithm for turbo codes, which alters the proportion of the extrinsic information and the received systematic observations in the soft output values of turbo decoding. In this way, turbo decoding can trace the changes of channel and obtain improved performance. Simulation results show that the proposed scheme can lower the error floor of turbo codes.
Cited by 1
Anders Gjendemsjø, Geir E. Øien, Henrik Holm · 2013
Cited by 0
CAI FENG · National University of Singapore · 2005
Master's
Cited by 0
Yang Xiao, Moon Ho Lee · 2006
Presented low-density parity-check (LDPC) decoding exhibits poor BER performance for multipath channel (MPC). To solve the problem, this paper proposes an adaptive LDPC decoding system, which composes two parts: adaptive equalizer and LDPC decoder. In this scheme, adaptive equalizer resorts to the training sequence transmitted from transmitter, removes the multipath interference (MPI), then LDPC decoder can work as in AWGN channel and exhibit good BER performance. Furthermore the decision on the exact decoding performance can be taken a posterior in the receiver and therefore requires no signaling whatsoever between sender and receiver. The simulations show that the proposed decoding system is feasible to resume the data to be translating in MPC
Cited by 2
Simon Geirnaert, Tom Francart, Alexander Bertrand · IEEE Journal of Biomedical and Health Informatics · 2021
When multiple speakers talk simultaneously, a hearing device cannot identify which of these speakers the listener intends to attend to. Auditory attention decoding (AAD) algorithms can provide this information by, for example, reconstructing the attended speech envelope from electroencephalography (EEG) signals. However, these stimulus reconstruction decoders are traditionally trained in a supervised manner, requiring a dedicated training stage during which the attended speaker is known. Pre-trained subject-independent decoders alleviate the need of having such a per-user training stage but perform substantially worse than supervised subject-specific decoders that are tailored to the user. This motivates the development of a new unsupervised self-adapting training/updating procedure for a subject-specific decoder, which iteratively improves itself on unlabeled EEG data using its own predicted labels. This iterative updating procedure enables a self-leveraging effect, of which we provide a mathematical analysis that reveals the underlying mechanics. The proposed unsupervised algorithm, starting from a random decoder, results in a decoder that outperforms a supervised subject-independent decoder. Starting from a subject-independent decoder, the unsupervised algorithm even closely approximates the performance of a supervised subject-specific decoder. The developed unsupervised AAD algorithm thus combines the two advantages of a supervised subject-specific and subject-independent decoder: it approximates the performance of the former while retaining the 'plug-and-play' character of the latter. As the proposed algorithm can be used to automatically adapt to new users, as well as over time when new EEG data is being recorded, it contributes to more practical neuro-steered hearing devices.
Cited by 22Open access
Yu Guo · Scholarship at UWindsor (University of Windsor) · 2014
A new adaptive filter is proposed for the turbo decoding on Rayleigh fading channels with noisy channel estimates. The extrinsic information generated from the turbo decoder has some priority information about the transmitted data bits, which can help us better understand the channel characters. By using the soft extrinsic information after each iteration of decoding, we re-estimate the channel and the minimum mean square error (m.m.s.e.) and further update the channel reliability factor and decision variables at each iteration. Simulations show that signal to noise (SNR) gain is improved by up to about 1dB at bit error probability of 3.5&times10-4.
Cited by 0Open access
Урніш, Віра Сергіївна, Кравчук, Сергій Олександрович · The Scientific Issues of Ternopil Volodymyr Hnatiuk National Pedagogical University Series pedagogy · 2021
This paper will consider the adaptive decode-and-forward protocol. Options for constructing a simple and complex decode-and-forward protocol are considered. The purpose of this protocol is to minimize the reception of messages with errors.
Cited by 0Open access
A.M. Breipohl, A Koschmann · 1964
the r e s u l t s of.a study of adaptive communication s y s t e m s i n which the "learning" is based on Bayes' Theorem.The learning is then Used t o guide adaptation whose purpose is t o optimize s y s t e m performance.Performance is evaluated by a m e a s u r e based on decision theory.The communication s y s t e m model used is a binary,sym.metricchanliel where, "r" - , .-denotes the probability of c o r r e c t transmission, and it is assumed that only the decoding is adaptable.Assuming a time invariant unknown r, i t is shown that the t r u e . . . ." value of r w i n be learned in the limit, if, and only if, t h e r e is some'redundancy in the message.It is also shown that the proposed adaptive system will converge t o the .best possible system if the t h e value of r is learned.: The speed of convergence is measured by the probability that the best possible . .decoding system is selected after n messages have been'received.T h i s m e a s u r e is calculated f o r a sample adaptive s y s t e m i n o r d e r t o show the effect of redundancy in the transmitted message, the number of m e s s a g e s r,eckived, and the t r u e value.ofr .
Cited by 0Open access
M. Sallout, Ebrahim A. Mattar · 2019
Cited by 0
M. Hossein Taghavi, P.H. Siegel · IEEE Transactions on Information Theory · 2008
<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> Detectability of failures of linear programming (LP) decoding and the potential for improvement by adding new constraints motivate the use of an adaptive approach in selecting the constraints for the underlying LP problem. In this paper, we make a first step in studying this method, and show that by starting from a simple LP problem and adaptively adding the necessary constraints, the complexity of LP decoding can be significantly reduced. In particular, we observe that with adaptive LP decoding, the sizes of the LP problems that need to be solved become practically independent of the density of the parity-check matrix. We further show that adaptively adding extra constraints, such as constraints based on redundant parity checks, can provide large gains in the performance. </para>
Cited by 120
Philippa A. Martin, D.P. Taylor · 2002
This paper develops an adaptive approach to iterative decoding of block turbo codes. The method developed in this paper allows the receiver to adapt to the conditions of the channel, the codes and the modulation scheme. The approach yields improved performance especially in early iterations, for a very small increase in complexity.
Cited by 15
Shehzaad Dhuliawala, Ilia Kulikov, Ping Yu et al. · arXiv (Cornell University) · 2024
During language model decoding, it is known that using higher temperature sampling gives more creative responses, while lower temperatures are more factually accurate. However, such models are commonly applied to general instruction following, which involves both creative and fact seeking tasks, using a single fixed temperature across all examples and tokens. In this work, we introduce Adaptive Decoding, a layer added to the model to select the sampling temperature dynamically at inference time, at either the token or example level, in order to optimize performance. To learn its parameters we introduce Latent Preference Optimization (LPO) a general approach to train discrete latent variables such as choices of temperature. Our method outperforms all fixed decoding temperatures across a range of tasks that require different temperatures, including UltraFeedback, Creative Story Writing, and GSM8K.
Cited by 1Open access
Agus Sudjianto, Kannan Venkataramanan · SSRN Electronic Journal · 2026
Cited by 0Open access