arXiv subjects

Ken R. Duffy

Publications and source records attributed to Ken R. Duffy.

At least 55 records · Page 3Linked to original sources

arXiv2022

Block turbo decoding with ORBGRAND

Guessing Random Additive Noise Decoding (GRAND) is a family of universal decoding algorithms suitable for decoding any moderate redundancy code of any length. We establish that, through the use of list decoding, soft-input variants of GRAND can replace the Chase algorithm as the component decoder in the turbo decoding of product codes. In addition to being able to decode arbitrary product codes, rather than just those with dedicated hard-input component code decoders, results show that ORBGRAND achieves a coding gain of up to 0.7dB over the Chase algorithm with same list size.

cs.IT↗

arXiv2022

Soft decoding without soft demapping with ORBGRAND

For spectral efficiency, higher order modulation symbols confer information on more than one bit. As soft detection forward error correction decoders assume the availability of information at binary granularity, however, soft demappers are required to compute per-bit reliabilities from complex-valued signals. Here we show that the recently introduced universal soft detection decoder ORBGRAND can be adapted to work with symbol-level soft information, obviating the need for energy expensive soft demapping. We establish that doing so reduces complexity while retaining the error correction performance achieved with the optimal demapper.

cs.IT↗

arXiv2022

Partial Encryption after Encoding for Security and Reliability in Data Systems

We consider the problem of secure and reliable communication over a noisy multipath network. Previous work considering a noiseless version of our problem proposed a hybrid universal network coding cryptosystem (HUNCC). By combining an information-theoretically secure encoder together with partial encryption, HUNCC is able to obtain security guarantees, even in the presence of an all-observing eavesdropper. In this paper, we propose a version of HUNCC for noisy channels (N-HUNCC). This modification requires four main novelties. First, we present a network coding construction which is jointly, individually secure and error-correcting. Second, we introduce a new security definition which is a computational analogue of individual security, which we call individual indistinguishability under chosen ciphertext attack (individual IND-CCA1), and show that NHUNCC satisfies it. Third, we present a noise based decoder for N-HUNCC, which permits the decoding of the encoded-thenencrypted data. Finally, we discuss how to select parameters for N-HUNCC and its error-correcting capabilities.

cs.IT↗

arXiv2022

Syfer: Neural Obfuscation for Private Data Release

Balancing privacy and predictive utility remains a central challenge for machine learning in healthcare. In this paper, we develop Syfer, a neural obfuscation method to protect against re-identification attacks. Syfer composes trained layers with random neural networks to encode the original data (e.g. X-rays) while maintaining the ability to predict diagnoses from the encoded data. The randomness in the encoder acts as the private key for the data owner. We quantify privacy as the number of attacker guesses required to re-identify a single image (guesswork). We propose a contrastive learning algorithm to estimate guesswork. We show empirically that differentially private methods, such as DP-Image, obtain privacy at a significant loss of utility. In contrast, Syfer achieves strong privacy while preserving utility. For example, X-ray classifiers built with DP-image, Syfer, and original data achieve average AUCs of 0.53, 0.78, and 0.86, respectively.

cs.LG↗

arXiv2021

Guessing random additive noise decoding with symbol reliability information (SRGRAND)

The design and implementation of error correcting codes has long been informed by two fundamental results: Shannon's 1948 capacity theorem, which established that long codes use noisy channels most efficiently; and Berlekamp, McEliece, and Van Tilborg's 1978 theorem on the NP-hardness of decoding linear codes. These results shifted focus away from creating code-independent decoders, but recent low-latency communication applications necessitate relatively short codes, providing motivation to reconsider the development of universal decoders. We introduce a scheme for employing binarized symbol soft information within Guessing Random Additive Noise Decoding, a universal hard detection decoder. We incorporate codebook-independent quantization of soft information to indicate demodulated symbols to be reliable or unreliable. We introduce two decoding algorithms: one identifies a conditional Maximum Likelihood (ML) decoding; the other either reports a conditional ML decoding or an error. For random codebooks, we present error exponents and asymptotic complexity, and show benefits over hard detection. As empirical illustrations, we compare performance with majority logic decoding of Reed-Muller codes, with Berlekamp-Massey decoding of Bose-Chaudhuri-Hocquenghem codes, with CA-SCL decoding of CA-Polar codes, and establish the performance of Random Linear Codes, which require a universal decoder and offer a broader palette of code sizes and rates than traditional codes.

cs.IT↗

arXiv2021

NeuraCrypt: Hiding Private Health Data via Random Neural Networks for Public Training

Balancing the needs of data privacy and predictive utility is a central challenge for machine learning in healthcare. In particular, privacy concerns have led to a dearth of public datasets, complicated the construction of multi-hospital cohorts and limited the utilization of external machine learning resources. To remedy this, new methods are required to enable data owners, such as hospitals, to share their datasets publicly, while preserving both patient privacy and modeling utility. We propose NeuraCrypt, a private encoding scheme based on random deep neural networks. NeuraCrypt encodes raw patient data using a randomly constructed neural network known only to the data-owner, and publishes both the encoded data and associated labels publicly. From a theoretical perspective, we demonstrate that sampling from a sufficiently rich family of encoding functions offers a well-defined and meaningful notion of privacy against a computationally unbounded adversary with full knowledge of the underlying data-distribution. We propose to approximate this family of encoding functions through random deep neural networks. Empirically, we demonstrate the robustness of our encoding to a suite of adversarial attacks and show that NeuraCrypt achieves competitive accuracy to non-private baselines on a variety of x-ray tasks. Moreover, we demonstrate that multiple hospitals, using independent private encoders, can collaborate to train improved x-ray models. Finally, we release a challenge dataset to encourage the development of new attacks on NeuraCrypt.

cs.CR↗

arXiv2021

CRC Codes as Error Correction Codes

CRC codes have long since been adopted in a vast range of applications. The established notion that they are suitable primarily for error detection can be set aside through use of the recently proposed Guessing Random Additive Noise Decoding (GRAND). Hard-detection (GRAND-SOS) and soft-detection (ORBGRAND) variants can decode any short, high-rate block code, making them suitable for error correction of CRC-coded data. When decoded with GRAND, short CRC codes have error correction capability that is at least as good as popular codes such as BCH codes, but with no restriction on either code length or rate. The state-of-the-art CA-Polar codes are concatenated CRC and Polar codes. For error correction, we find that the CRC is a better short code than either Polar or CA-Polar codes. Moreover, the standard CA-SCL decoder only uses the CRC for error detection and therefore suffers severe performance degradation in short, high rate settings when compared with the performance GRAND provides, which uses all of the CA-Polar bits for error correction. Using GRAND, existing systems can be upgraded from error detection to low-latency error correction without re-engineering the encoder, and additional applications of CRCs can be found in IoT, Ultra-Reliable Low Latency Communication (URLLC), and beyond. The universality of GRAND, its ready parallelized implementation in hardware, and the good performance of CRC as codes make their combination a viable solution for low-latency applications.

cs.IT↗

arXiv2020

Keep the bursts and ditch the interleavers

To facilitate applications in IoT, 5G, and beyond, there is an engineering need to enable high-rate, low-latency communications. Errors in physical channels typically arrive in clumps, but most decoders are designed assuming that channels are memoryless. As a result, communication networks rely on interleaving over tens of thousands of bits so that channel conditions match decoder assumptions. Even for short high rate codes, awaiting sufficient data to interleave at the sender and de-interleave at the receiver is a significant source of unwanted latency. Using existing decoders with non-interleaved channels causes a degradation in block error rate performance owing to mismatch between the decoder's channel model and true channel behaviour. Through further development of the recently proposed Guessing Random Additive Noise Decoding (GRAND) algorithm, which we call GRAND-MO for GRAND Markov Order, here we establish that by abandoning interleaving and embracing bursty noise, low-latency, short-code, high-rate communication is possible with block error rates that outperform their interleaved counterparts by a substantial margin. Moreover, while most decoders are twinned to a specific code-book structure, GRAND-MO can decode any code. Using this property, we establish that certain well-known structured codes are ill-suited for use in bursty channels, but Random Linear Codes (RLCs) are robust to correlated noise. This work suggests that the use of RLCs with GRAND-MO is a good candidate for applications requiring high throughput with low latency.

cs.IT↗

arXiv2020

Noise Recycling

We introduce Noise Recycling, a method that enhances decoding performance of channels subject to correlated noise without joint decoding. The method can be used with any combination of codes, code-rates and decoding techniques. In the approach, a continuous realization of noise is estimated from a lead channel by subtracting its decoded output from its received signal. This estimate is then used to improve the accuracy of decoding of an orthogonal channel that is experiencing correlated noise. In this design, channels aid each other only through the provision of noise estimates post-decoding. In a Gauss-Markov model of correlated noise, we constructive establish that noise recycling employing a simple successive order enables higher rates than not recycling noise. Simulations illustrate noise recycling can be employed with any code and decoder, and that noise recycling shows Block Error Rate (BLER) benefits when applying the same predetermined order as used to enhance the rate region. Finally, for short codes we establish that an additional BLER improvement is possible through noise recycling with racing, where the lead channel is not pre-determined, but is chosen on the fly based on which decoder completes first.

cs.IT↗

arXiv2020

Ordered Reliability Bits Guessing Random Additive Noise Decoding

Modern applications are driving demand for ultra-reliable low-latency communications, rekindling interest in the performance of short, high-rate error correcting codes. To that end, here we introduce a soft-detection variant of Guessing Random Additive Noise Decoding (GRAND) called Ordered Reliability Bits GRAND that can decode any short, high-rate block-code. For a code of $n$ bits, it avails of no more than $\lceil\log_2(n)\rceil$ bits of code-book-independent quantized soft detection information per received bit to determine an accurate decoding while retaining the original algorithm's suitability for a highly parallelized implementation in hardware. ORBGRAND is shown to provide similar block error performance for codes of distinct classes (BCH, CA-Polar and RLC) with low complexity, while providing better block error rate performance than CA-SCL, a state of the art soft detection CA-Polar decoder.

cs.IT↗

arXiv2020

Discrete convolution statistic for hypothesis testing

The question of testing for equality in distribution between two linear models, each consisting of sums of distinct discrete independent random variables with unequal numbers of observations, has emerged from the biological research. In this case, the computation of classical $χ^2$ statistics, which would not include all observations, results in loss of power, especially when sample sizes are small. Here, as an alternative that uses all data, the nonparametric maximum likelihood estimator for the distribution of sum of discrete and independent random variables, which we call the convolution statistic, is proposed and its limiting normal covariance matrix determined. To challenge null hypotheses about the distribution of this sum, the generalized Wald's method is applied to define a testing statistic whose distribution is asymptotic to a $χ^2$ with as many degrees of freedom as the rank of such covariance matrix. Rank analysis also reveals a connection with the roots of the probability generating functions associated to the addend variables of the linear models. A simulation study is performed to compare the convolution test with Pearson's $χ^2$, and to provide usage guidelines.

math.ST↗

arXiv2020

Noise Recycling

We introduce Noise Recycling, a method that substantially enhances decoding performance of orthogonal channels subject to correlated noise without the need for joint encoding or decoding. The method can be used with any combination of codes, code-rates and decoding techniques. In the approach, a continuous realization of noise is estimated from a lead channel by subtracting its decoded output from its received signal. The estimate is recycled to reduce the Signal to Noise Ratio (SNR) of an orthogonal channel that is experiencing correlated noise and so improve the accuracy of its decoding. In this design, channels only aid each other only through the provision of noise estimates post-decoding. For a system with arbitrary noise correlation between orthogonal channels experiencing potentially distinct conditions, we introduce an algorithm that determines a static decoding order that maximizes total effective SNR. We prove that this solution results in higher effective SNR than independent decoding, which in turn leads to a larger rate region. We derive upper and lower bounds on the capacity of any sequential decoding of orthogonal channels with correlated noise where the encoders are independent and show that those bounds are almost tight. We numerically compare the upper bound with the capacity of jointly Gaussian noise channel with joint encoding and decoding, showing that they match. Simulation results illustrate that Noise Recycling can be employed with any combination of codes and decoders, and that it gives significant Block Error Rate (BLER) benefits when applying the static predetermined order used to enhance the rate region. We further establish that an additional BLER improvement is possible through Dynamic Noise Recycling, where the lead channel is not pre-determined but is chosen on-the-fly based on which decoder provides the most confident decoding.

cs.IT↗

arXiv2020

Inferring differentiation order in adaptive immune responses from population level data

A hallmark of the adaptive immune response is the proliferation of pathogen-specific lymphocytes that leave in their wake a long lived population of cells that provide lasting immunity. A subject of ongoing investigation is when during an adaptive immune response those memory cells are produced. In two ground-breaking studies, Buchholz et al. (Science, 2013) and Gerlach et al. (Science, 2013) employed experimental methods that allowed identification of offspring from individual lymphocytes in vivo, which we call clonal data, at a single time point. Through the development, application and fitting of a mathematical model, Buchholz et al. (Science, 2013) concluded that, if memory is produced during the expansion phase, memory cell precursors are made before the effector cells that clear the original pathogen. We sought to determine the general validity and power of the modeling approach introduced in Buchholz et al. (Science, 2013) for quickly evaluating differentiation networks by adapting it to make it suitable for drawing inferences from more readily available non-clonal phenotypic proportion time-courses. We first established the method drew consistent deductions when fit to the non-clonal data in Buchholz et al. (Science, 2013) itself. We fit a variant of the model to data reported in Badovinac et al. (J. Immun., 2007), Schlub et al. (Immun. & Cell Bio., 2010), and Kinjo et al. (Nature Commun., 2015) with necessary simplifications to match different reported data in these papers. The deduction from the model was consistent with that in Buchholz et al. (Science, 2013), albeit with questionable parameterizations. An alternative possibility, supported by the data in Kinjo et al. (Nature Commun., 2015), is that memory precursors are created after the expansion phase, which is a deduction not possible from the mathematical methods provided in Buchholz et al. (Science, 2013).

q-bio.CB↗

arXiv2020

Privacy with Estimation Guarantees

We study the central problem in data privacy: how to share data with an analyst while providing both privacy and utility guarantees to the user that owns the data. In this setting, we present an estimation-theoretic analysis of the privacy-utility trade-off (PUT). Here, an analyst is allowed to reconstruct (in a mean-squared error sense) certain functions of the data (utility), while other private functions should not be reconstructed with distortion below a certain threshold (privacy). We demonstrate how chi-square information captures the fundamental PUT in this case and provide bounds for the best PUT. We propose a convex program to compute privacy-assuring mappings when the functions to be disclosed and hidden are known a priori and the data distribution is known. We derive lower bounds on the minimum mean-squared error of estimating a target function from the disclosed data and evaluate the robustness of our approach when an empirical distribution is used to compute the privacy-assuring mappings instead of the true data distribution. We illustrate the proposed approach through two numerical experiments.

cs.IT↗

arXiv2020

Soft Maximum Likelihood Decoding using GRAND

Maximum Likelihood (ML) decoding of forward error correction codes is known to be optimally accurate, but is not used in practice as it proves too challenging to efficiently implement. Here we introduce a ML decoder called SGRAND, which is a development of a previously described hard detection ML decoder called GRAND, that fully avails of soft detection information and is suitable for use with any arbitrary high-rate, short-length block code. We assess SGRAND's performance on CRC-aided Polar (CA-Polar) codes, which will be used for all control channel communication in 5G NR, comparing its accuracy with CRC-Aided Successive Cancellation List decoding (CA-SCL), a state-of-the-art soft-information decoder specific to CA-Polar codes.

cs.IT↗

arXiv2019

Sample Path Properties of the Average Generation of a Bellman-Harris Process

Motivated by a recently proposed design for a DNA coded randomised algorithm that enables inference of the average generation of a collection of cells descendent from a common progenitor, here we establish strong convergence properties for the average generation of a super-critical Bellman-Harris process. We further extend those results to a two-type Bellman-Harris process where one type can give rise to the other, but not vice versa. These results further affirm the estimation method's potential utility by establishing its long run accuracy on individual sample-paths, and significantly expanding its remit to encompass cellular development that gives rise to differentiated offspring with distinct population dynamics.

math.PR↗

arXiv2019

Capacity-achieving Guessing Random Additive Noise Decoding (GRAND)

We introduce a new algorithm for realizing Maximum Likelihood (ML) decoding in discrete channels with or without memory. In it, the receiver rank orders noise sequences from most likely to least likely. Subtracting noise from the received signal in that order, the first instance that results in a member of the code-book is the ML decoding. We name this algorithm GRAND for Guessing Random Additive Noise Decoding. We establish that GRAND is capacity-achieving when used with random code-books. For rates below capacity we identify error exponents, and for rates beyond capacity we identify success exponents. We determine the scheme's complexity in terms of the number of computations the receiver performs. For rates beyond capacity, this reveals thresholds for the number of guesses by which if a member of the code-book is identified it is likely to be the transmitted code-word. We introduce an approximate ML decoding scheme where the receiver abandons the search after a fixed number of queries, an approach we dub GRANDAB, for GRAND with ABandonment. While not an ML decoder, we establish that the algorithm GRANDAB is also capacity-achieving for an appropriate choice of abandonment threshold, and characterize its complexity, error and success exponents. Worked examples are presented for Markovian noise that indicate these decoding schemes substantially out-perform the brute force decoding approach.

cs.IT↗

arXiv2019

The variance of the average depth of a pure birth process converges to 7

If trees are constructed from a pure birth process and one defines the depth of a leaf to be the number of edges to its root, it is known that the variance in the depth of a randomly selected leaf of a randomly selected tree grows linearly in time. In this letter, we instead consider the variance of the average depth of leaves within each individual tree, establishing that, in contrast, it converges to a constant, $7$. This result indicates that while the variance in leaf depths amongst the ensemble of pure birth processes undergoes large fluctuations, the average depth across individual trees is much more consistent.

math.PR↗