Search

Okta logs are used today to detect cybersecurity events using various rule-based models with restricted look back periods. These functions have limitations, such as a limited retrospective analysis, a predefined rule set, and susceptibility to generating false positives. To address this, we adopt unsupervised techniques, specifically employing autoencoders. To properly use an autoencoder, we need to transform and simplify the complexity of the log data we receive from our users. This transformed and filtered data is then fed into the autoencoder, and the output is evaluated., Comment: 11 pages, 3 tables, 8 figures, Databricks AI Summit 2024

Quantum low-density parity-check (qLDPC) codes are an important component in the quest for quantum fault tolerance. Dramatic recent progress on qLDPC codes has led to constructions which are asymptotically good, and which admit linear-time decoders to correct errors affecting a constant fraction of codeword qubits. These constructions, while theoretically explicit, rely on inner codes with strong properties only shown to exist by probabilistic arguments, resulting in lengths that are too large to be practically relevant. In practice, the surface/toric codes, which are the product of two repetition codes, are still often the qLDPC codes of choice. A previous construction based on the lifted product of an expander-based classical LDPC code with a repetition code (Panteleev & Kalachev, 2020) achieved a near-linear distance (of $\Omega(N/\log N)$ where $N$ is the number of codeword qubits), and avoids the need for such intractable inner codes. Our main result is an efficient decoding algorithm for these codes that corrects $\Theta(N/\log N)$ adversarial errors. En route, we give such an algorithm for the hypergraph product version these codes, which have weaker $\Theta(\sqrt{N})$ distance (but are simpler). Our decoding algorithms leverage the fact that the codes we consider are quasi-cyclic, meaning that they respect a cyclic group symmetry. Since the repetition code is not based on expanders, previous approaches to decoding expander-based qLDPC codes, which typically worked by greedily flipping code bits to reduce some potential function, do not apply in our setting. Instead, we reduce our decoding problem (in a black-box manner) to that of decoding classical expander-based LDPC codes under noisy parity-check syndromes. For completeness, we also include a treatment of such classical noisy-syndrome decoding that is sufficient for our application to the quantum setting.

The seminal work of Bencz\'ur and Karger demonstrated cut sparsifiers of near-linear size, with several applications throughout theoretical computer science. Subsequent extensions have yielded sparsifiers for hypergraph cuts and more recently linear codes over Abelian groups. A decade ago, Kogan and Krauthgamer asked about the sparsifiability of arbitrary constraint satisfaction problems (CSPs). For this question, a trivial lower bound is the size of a non-redundant CSP instance, which admits, for each constraint, an assignment satisfying only that constraint (so that no constraint can be dropped by the sparsifier). For graph cuts, spanning trees are non-redundant instances. Our main result is that redundant clauses are sufficient for sparsification: for any CSP predicate R, every unweighted instance of CSP(R) has a sparsifier of size at most its non-redundancy (up to polylog factors). For weighted instances, we similarly pin down the sparsifiability to the so-called chain length of the predicate. These results precisely determine the extent to which any CSP can be sparsified. A key technical ingredient in our work is a novel application of the entropy method from Gilmer's recent breakthrough on the union-closed sets conjecture. As an immediate consequence of our main theorem, a number of results in the non-redundancy literature immediately extend to CSP sparsification. We also contribute new techniques for understanding the non-redundancy of CSP predicates. In particular, we give an explicit family of predicates whose non-redundancy roughly corresponds to the structure of matching vector families in coding theory. By adapting methods from the matching vector codes literature, we are able to construct an explicit predicate whose non-redundancy lies between $\Omega(n^{1.5})$ and $\widetilde{O}(n^{1.6})$, the first example with a provably non-integral exponent., Comment: 66 pages

We present new constructions of quantum codes of linear or close-to-linear distance and dimension with low-weight stabilizers. Only a few constructions of such codes were previously known, and were primarily based on a specific operation from homological algebra, namely the balanced product. In contrast, our constructions are based on a more basic and widely used product, namely the homological product (i.e. the tensor product of chain complexes). Our results help address the natural question: When do homological products preserve good code distance? Our first main result constructs asymptotically good $[[N,\Theta(N),\Theta(N)]]$ quantum codes with small polynomial stabilizer weight from homological products of codes with a property called product-expansion. This notion was recently introduced and used to bound the distance of balanced product quantum codes; we apply it instead to homological products. For every $\epsilon>0$, our second main result constructs close-to-linear distance $[[N,N^{1-\epsilon},N^{1-\epsilon}]]$ (subsystem) quantum LDPC codes with constant stabilizer weight from iterated homological products of a constant-sized quantum locally testable code. The key insight here is that by using subsystem codes (but still with constant-weight stabilizers), we can circumvent a particular obstruction that limited the distance of many prior product code constructions to at most $\tilde{O}(\sqrt{N})$.

Cervical cancer remains the fourth most common malignancy amongst women worldwide.1 Concurrent chemoradiotherapy (CRT) serves as the mainstay definitive treatment regimen for locally advanced cervical cancers and includes external beam radiation followed by brachytherapy.2 Integral to radiotherapy treatment planning is the routine contouring of both the target tumor at the level of the cervix, associated gynecologic anatomy and the adjacent organs at risk (OARs). However, manual contouring of these structures is both time and labor intensive and associated with known interobserver variability that can impact treatment outcomes. While multiple tools have been developed to automatically segment OARs and the high-risk clinical tumor volume (HR-CTV) using computed tomography (CT) images,3,4,5,6 the development of deep learning-based tumor segmentation tools using routine T2-weighted (T2w) magnetic resonance imaging (MRI) addresses an unmet clinical need to improve the routine contouring of both anatomical structures and cervical cancers, thereby increasing quality and consistency of radiotherapy planning. This work applied a novel deep-learning model (PocketNet) to segment the cervix, vagina, uterus, and tumor(s) on T2w MRI. The performance of the PocketNet architecture was evaluated, when trained on data via 5-fold cross validation. PocketNet achieved a mean Dice-Sorensen similarity coefficient (DSC) exceeding 70% for tumor segmentation and 80% for organ segmentation. These results suggest that PocketNet is robust to variations in contrast protocols, providing reliable segmentation of the ROIs.

Introduced in [CG24], pseudorandom error-correcting codes (PRCs) are a new cryptographic primitive with applications in watermarking generative AI models. These are codes where a collection of polynomially many codewords is computationally indistinguishable from random, except to individuals with the decoding key. In this work, we examine the assumptions under which PRCs with robustness to a constant error rate exist. 1. We show that if both the planted hyperloop assumption introduced in [BKR23] and security of a version of Goldreich's PRG hold, then there exist public-key PRCs for which no efficient adversary can distinguish a polynomial number of codewords from random with better than $o(1)$ advantage. 2. We revisit the construction of [CG24] and show that it can be based on a wider range of assumptions than presented in [CG24]. To do this, we introduce a weakened version of the planted XOR assumption which we call the weak planted XOR assumption and which may be of independent interest. 3. We initiate the study of PRCs which are secure against space-bounded adversaries. We show how to construct secret-key PRCs of length $O(n)$ which are $\textit{unconditionally}$ indistinguishable from random by $\text{poly}(n)$ time, $O(n^{1.5-\varepsilon})$ space adversaries., Comment: 37 pages, 1 figure

As a possible implementation of data storage using DNA, multiple strands of DNA are stored in a liquid container so that, in the future, they can be read by an array of DNA readers in parallel. These readers will sample the strands with replacement to produce a random number of noisy reads for each strand. An essential component of such a data storage system is how to reconstruct data out of these unsorted, repetitive, and noisy reads. It is known that if a single read can be modeled by a substitution channel $W$, then the overall capacity can be expressed by the "Poisson-ization" of $W$. In this paper, we lay down a rateless code along each strand to encode its index; we then lay down a capacity-achieving block code at the same position across all strands to protect data. That weaves a low-complexity coding scheme that achieves DNA's capacity., Comment: 18 pages, 5 figures

We construct quantum codes that support transversal $CCZ$ gates over qudits of arbitrary prime power dimension $q$ (including $q=2$) such that the code dimension and distance grow linearly in the block length. The only previously known construction with such linear dimension and distance required a growing alphabet size $q$ (Krishna & Tillich, 2019). Our codes imply protocols for magic state distillation with overhead exponent $\gamma=\log(n/k)/\log(d)\rightarrow 0$ as the block length $n\rightarrow\infty$, where $k$ and $d$ denote the code dimension and distance respectively. It was previously an open question to obtain such a protocol with a contant alphabet size $q$. We construct our codes by combining two modular components, namely, (i) a transformation from classical codes satisfying certain properties to quantum codes supporting transversal $CCZ$ gates, and (ii) a concatenation scheme for reducing the alphabet size of codes supporting transversal $CCZ$ gates. For this scheme we introduce a quantum analogue of multiplication-friendly codes, which provide a way to express multiplication over a field in terms of a subfield. We obtain our asymptotically good construction by instantiating (i) with algebraic-geometric codes, and applying a constant number of iterations of (ii). We also give an alternative construction with nearly asymptotically good parameters ($k,d=n/2^{O(\log^*n)}$) by instantiating (i) with Reed-Solomon codes and then performing a superconstant number of iterations of (ii).

We present a systematic moment expansion for solving the semiclassical Boltzmann equations for electroweak baryogenesis. The expansion is developed in powers of adiabatic coordinate velocity, and it is used for computing the CP-violating seed asymmetry at the front of the phase transition wall, that sources the eventual baryon asymmetry of the universe (BAU). We implement the method in a benchmark model, with a CP-violating mass arising from a dimension-5 operator coupling the fermion to a singlet scalar field. We find that the higher moment calculations yield a BAU that can significantly differ from the commonly used two-moment approximation. We discuss in detail the underlying approximations in the moment method and propose a new truncation scheme for the expansion, that appears to give more numerically robust results than the previous schemes., Comment: 24 pages, 8 figures

In order to communicate a message over a noisy channel, a sender (Alice) uses an error-correcting code to encode her message $x$ into a codeword. The receiver (Bob) decodes it correctly whenever there is at most a small constant fraction of adversarial error in the transmitted codeword. This work investigates the setting where Bob is computationally bounded. Specifically, Bob receives the message as a stream and must process it and write $x$ in order to a write-only tape while using low (say polylogarithmic) space. We show three basic results about this setting, which are informally as follows: (1) There is a stream decodable code of near-quadratic length. (2) There is no stream decodable code of sub-quadratic length. (3) If Bob need only compute a private linear function of the input bits, instead of writing them all to the output tape, there is a stream decodable code of near-linear length.

To ensure differential privacy, one can reveal an integer fuzzily in two ways: (a) add some Laplace noise to the integer, or (b) encode the integer as a binary string and add iid BSC noise. The former is simple and natural while the latter is flexible and affordable, especially when one wants to reveal a sparse vector of integers. In this paper, we propose an implementation of (b) that achieves the capacity of the BSC with positive error exponents. Our implementation adds error-correcting functionality to Gray codes by mimicking how software updates back up the files that are getting updated ("coded Gray code"). In contrast, the old implementation of (b) interpolates between codewords of a black-box error-correcting code ("Grayed code")., Comment: 10 pages, 3 figures

The problem of efficiently coloring $3$-colorable graphs with few colors has received much attention on both the algorithmic and inapproximability fronts. We consider exponential time approximations, in which given a parameter $r$, we aim to develop an $r$-approximation algorithm with the best possible runtime, providing a tradeoff between runtime and approximation ratio. In this vein, an algorithm to $O(n^\varepsilon)$-color a 3-colorable graphs in time $2^{\Theta(n^{1-2\varepsilon}\log(n))}$ is given in (Atserias and Dalmau, SODA 2022.) We build on tools developed in (Bansal et al., Algorithmic, 2019) to obtain an algorithm to color $3$-colorable graphs with $O(r)$ colors in $\exp\left(\tilde{O}\left(\frac {n\log^{11/2}r} {r^3}\right)\right)$ time, asymptotically improving upon the bound given by Atserias and Dalmau.

We study the biological evolution of low-latency natural neural networks for short-term survival, and its parallels in the development of low latency high-performance Central Processing Unit in computer design and architecture. The necessity of accurate high-quality display of motion picture led to the special processing units known as the GPU, just as how special visual cortex regions of animals produced such low-latency computational capacity. The human brain, especially considered as nothing but a scaled-up version of a primate brain evolved in response to genomic bottleneck, producing a brain that is trainable and prunable by society, and as a further extension, invents language, writing and storage of narratives displaced in time and space. We conclude that this modern digital invention of social media and the archived collective common corpus has further evolved from just simple CPU-based low-latency fast retrieval to high-throughput parallel processing of data using GPUs to train Attention based Deep Learning Neural Networks producing Generative AI with aspects like toxicity, bias, memorization, hallucination, with intriguing close parallels in humans and their society. We show how this paves the way for constructive approaches to eliminating such drawbacks from human society and its proxy and collective large-scale mirror, the Generative AI of the LLMs.

Successive cancellation list (SCL) decoders of polar codes excel in practical performance but pose challenges for theoretical analysis. Existing works either limit their scope to erasure channels or address general channels without taking advantage of soft information. In this paper, we propose the "successive cancellation sampling" (SCS) decoder. SCS hires iid "agents" to sample codewords using posterior probabilities. This makes it fully parallel and amenable for some theoretical analysis. As an example, when comparing SCS with $a$ agents to any list decoder with list size $\ell$, we can prove that the error probability of the former is at most $\ell/ae$ more than that of the latter. In this paper, we also describe how to adjust the "temperature" of agents. Warmer agents are less likely to sample the same codewords and hence can further reduce error probability., Comment: 6 pages, 4 figures, ISIT 2024

Group testing (GT) is the art of identifying binary signals and the marketplace for exchanging new ideas for related fields such as unique-element counting, compressed sensing, traitor tracing, and geno-typing. A GT scheme can be nonadaptive or adaptive; the latter is preferred when latency is ess of an issue. To construct adaptive GT schemes, a popular strategy is to spend the majority of tests in the first few rounds to gain as much information as possible, and uses later rounds to refine details. In this paper, we propose a transparent strategy called "isolate and then identify" (I@I). In the first few rounds, I@I divides the population into teams until every team contains at most one sick person. Then, in the last round, I@I identifies the sick person in each team. Performance-wise, I@I is the first GT scheme that achieves the optimal coefficient $1/$capacity$(Z)$ for the $k \log_2 (n/k)$ term in the number of tests when $Z$ is a generic channel corrupting the test outcomes. I@I follows a modular methodology whereby the isolating part and the identification part can be optimized separately., Comment: 6 pages, 1 figure, ISIT 2024

We modify Cheraghchi-Nakos [CN20] and Price-Scarlett's [PS20] fast binary splitting approach to nonadaptive group testing. We show that, to identify a uniformly random subset of $k$ infected persons among a population of $n$, it takes only $\ln(2 - 4\varepsilon) ^{-2} k \ln n$ tests and decoding complexity $O(\varepsilon^{-2} k \ln n)$, for any small $\varepsilon > 0$, with vanishing error probability. In works prior to ours, only two types of group testing schemes exist. Those that use $\ln(2)^{-2} k \ln n$ or fewer tests require linear-in-$n$ complexity, sometimes even polynomial in $n$; those that enjoy sub-$n$ complexity employ $O(k \ln n)$ tests, where the big-$O$ scalar is implicit, presumably greater than $\ln(2)^{-2}$. We almost achieve the best of both worlds, namely, the almost-$\ln(2)^{-2}$ scalar and the sub-$n$ decoding complexity. How much further one can reduce the scalar $\ln(2)^{-2}$ remains an open problem., Comment: 6 pages, 3 figures, ISIT 2023

Polar codes that approach capacity at a near-optimal speed, namely with scaling exponents close to $2$, have been shown possible for $q$-ary erasure channels (Pfister and Urbanke), the BEC (Fazeli, Hassani, Mondelli, and Vardy), all BMS channels (Guruswami, Riazanov, and Ye), and all DMCs (Wang and Duursma). There is, nevertheless, a subtlety separating the last two papers from the first two, namely the usage of multiple dynamic kernels in the polarization process, which leads to increased complexity and fewer opportunities to hardware-accelerate. This paper clarifies this subtlety, providing a trade-off between the number of kernels in the construction and the scaling exponent. We show that the number of kernels can be bounded by $O(\ell^{3/\mu-1})$ where $\mu$ is the targeted scaling exponent and $\ell$ is the kernel size. In particular, if one settles for scaling exponent approaching $3$, a single kernel suffices, and to approach the optimal scaling exponent of $2$, about $O(\sqrt{\ell})$ kernels suffice., Comment: 6 pages, 2 figures, ISIT 2023

Formulating mitigation strategies is one of the main aspect in the dynamical study of extreme events. Apart from the effective control, easy implementation of the devised tool should also be given importance. In this work, we analyze the mitigation of extreme events in a coupled FitzHugh-Nagumo (FHN) neuron model utilizing an easily implementable constant bias analogous to a constant DC stimulant. We report the route through which the extreme events gets mitigated in $Two$, $Three$ and $N-$coupled FHN systems. In all the three cases, extreme events in the observable $\bar{x}$ gets suppressed. We confirm our results with the probability distribution function of peaks, $d_{max}$ plot and probability plots. Here $d_{max}$ is a measure of number of standard deviations that crosses the average amplitude corresponding to $\bar{x}_{max}$. Interestingly, we found that constant bias suppresses the extreme events without changing the collective frequency of the system., Comment: 19 pages, 25 figures

Biomedical data is growing exponentially, and managing it is increasingly challenging. While Findable, Accessible, Interoperable and Reusable (FAIR) data principles provide guidance, their adoption has proven difficult, especially in larger enterprises like pharmaceutical companies. In this manuscript, we describe how we leverage an Ontology-Based Data Management (OBDM) strategy for digital transformation in Novo Nordisk Research & Early Development. Here, we include both our technical blueprint and our approach for organizational change management. We further discuss how such an OBDM ecosystem plays a pivotal role in the organizations digital aspirations for data federation and discovery fuelled by artificial intelligence. Our aim for this paper is to share the lessons learned in order to foster dialogue with parties navigating similar waters while collectively advancing the efforts in the fields of data management, semantics and data driven drug discovery., Comment: 14 pages, 2 figures

We consider the task of certifying that a random $d$-dimensional subspace $X$ in $\mathbb{R}^n$ is well-spread - every vector $x \in X$ satisfies $c\sqrt{n} \|x\|_2 \leq \|x\|_1 \leq \sqrt{n}\|x\|_2$. In a seminal work, Barak et. al. showed a polynomial-time certification algorithm when $d \leq O(\sqrt{n})$. On the other hand, when $d \gg \sqrt{n}$, the certification task is information-theoretically possible but there is evidence that it is computationally hard [MW21,Cd22], a phenomenon known as the information-computation gap. In this paper, we give subexponential-time certification algorithms in the $d \gg \sqrt{n}$ regime. Our algorithm runs in time $\exp(\widetilde{O}(n^{\varepsilon}))$ when $d \leq \widetilde{O}(n^{(1+\varepsilon)/2})$, establishing a smooth trade-off between runtime and the dimension. Our techniques naturally extend to the related planted problem, where the task is to recover a sparse vector planted in a random subspace. Our algorithm achieves the same runtime and dimension trade-off for this task., Comment: 32 pages, 2 Figures