Maximum Information Research Articles

Community detection in hypergraphs via mutual information maximization

The hypergraph community detection problem seeks to identify groups of related vertices in hypergraph data. We propose an information-theoretic hypergraph community detection algorithm which compresses the observed data in terms of community labels and community-edge intersections. This algorithm can also be viewed as maximum-likelihood inference in a degree-corrected microcanonical stochastic blockmodel. We perform the compression/inference step via simulated annealing. Unlike several recent algorithms based on canonical models, our microcanonical algorithm does not require inference of statistical parameters such as vertex degrees or pairwise group connection rates. Through synthetic experiments, we find that our algorithm succeeds down to recently-conjectured thresholds for sparse random hypergraphs. We also find competitive performance in cluster recovery tasks on several hypergraph data sets.

Abstract 2299: More information from limited DNA: simultaneous measurement of genetics, 5hmC and 5mC in cell-free DNA

Abstract Liquid biopsy for profiling of cell free DNA (cfDNA) in blood holds huge promise to transform how we experience and manage cancer by early detection and identification of residual disease and subtype. While early work in liquid biopsy focused on the identification of actionable somatic variations at specific loci, the past decade has seen an expansion into non-genetic features, notably methylation. 5-methylcytosine (5-mC) profiles of cancer are differential from non-cancer at many more loci and so provide a stronger signal. Moreover, recent research has suggested that 5-hydroxymethylcytosine (5-hmC) profiles in cfDNA can be a marker for early cancer. However, a standard blood draw yields an average of only 10ng of cfDNA, presenting the dilemma of how to use limited sample to obtain maximum information. Existing methods for measuring methylation using next-generation sequencing cannot distinguish between 5-mC and 5-hmC and are limited in their ability to detect mutations, with approaches that measure both 5-mC and 5-hmC requiring two separate workflows. Hence, to measure complete genetics, 5-mC and 5-hmC using existing technologies requires three separate workflows, each of which require separate DNA input. We will present a technology which sequences at single base resolution the complete genetic sequence of input DNA fragments integrated with the modification status (unmodified, 5-mC or 5-hmC) for each CpG from low nanogram input quantities of DNA. A hairpin construct is used to create a copy of the original DNA strand. Enzymatic conversion followed by next-generation sequencing enables coupled decoding of bases across the original and copy strand, uniquely reporting one of A, C, G, T, 5-mC, or 5-hmC for each position in the input DNA fragment. Using this technology, we generated whole genome 6-letter data (measuring A, C, G, T, 5-mC, and 5-hmC at single-base resolution across the genome) on cell-free DNA extracted from plasma of healthy volunteers and patients with colorectal cancer at various stages of progression. We demonstrate how the technology can be used to compare 5-mC and 5-hmC methylomic profiles, genomics, and fragmentomics, across different stages of colorectal cancer. In particular, we show how disease progression can be marked by changes in 5-mC and 5-hmC across several loci. We propose that the ability to measure 5-mC and 5-hmC at high accuracy and single base resolution, alongside genomic and fragmentomic profiles, from a limited quantity of DNA will enable greater insight into the ctDNA in plasma and help advance the field of liquid biopsy towards fulfilling its promise. Citation Format: Tom J. Charlesworth, Fabio Puddu, Robert Crawford, Annelie Johansson, Ermira Lleshi, Aurelie Modat, Jamie Scotcher, Michael Wilson, Nicholas Harding, Jean Teyssandier, Jens Fullgrabe, Walraj Gosal, Paidi Creed. More information from limited DNA: simultaneous measurement of genetics, 5hmC and 5mC in cell-free DNA [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2299.

Impacts of locust feeding on interspecific relationships and niche of the major plants in Inner Mongolia grasslands

Oedaleus decorus asiaticus (O. asiaticus), the dominant species of steppe locust, plays a crucial role in driving the dynamic changes of the grassland ecosystem. However, there are a limited number of studies on the impact of locust feeding on the relationship of the major plants in the species community. This study has employed the Shannon-Wiener index, Pianka index, and the maximum information coefficient (MIC) to assess the impact of O. asiaticus feeding on the community structure and interspecific relationships in the grasslands of Inner Mongolia. The chosen indices measured niche breadth, niche overlap, and interspecies correlation. The topological network structure between the main species was further analyzed. The results indicated that: 1) locust gnawing expands the niche width of major species and increases the niche overlap index of food-loving species; 2) the feeding of locusts reduces the overall correlation between species but increases the interspecific relatedness between food-loving species and other species. 3) the number of links in the species topological network structure is reduced due to locust gnawing.

Optimal individual and collective measurements for nonorthogonal quantum key distribution signals

We consider how the theory of optimal quantum measurements determines the maximum information available to the receiving party of a quantum key distribution (QKD) system employing linearly independent but nonorthogonal quantum states. Such a setting is characteristic of several practical QKD protocols. Due to nonorthogonality, the receiver is not able to discriminate unambiguously between the signals. To understand the fundamental limits that this imposes, the quantity of interest is the maximum mutual information between the transmitter (Alice) and the receiver, whether legitimate (Bob) or an eavesdropper (Eve). To find the optimal measurement—taken individually or collectively—we use a framework based on operator algebra and general results derived from singular-value decomposition, achieving optimal solutions for von Neumann measurements and positive operator-valued measures (POVMs). The formal proof and quantitative analysis elaborated for two signals allow us to conclude that optimal von Neumann measurements are uniquely defined and provide a higher information gain compared to POVMs. Interestingly, collective measurements not only do not provide additional information gain with respect to individual ones but also suffer from a gain reduction in the case of POVMs. Published by the American Physical Society 2024

Boosting likelihood learning with event reweighting

Extracting maximal information from experimental data requires access to the likelihood function, which however is never directly available for complex experiments like those performed at high energy colliders. Theoretical predictions are obtained in this context by Monte Carlo events, which do furnish an accurate but abstract and implicit representation of the likelihood. Strategies based on statistical learning are currently being developed to infer the likelihood function explicitly by training a continuous-output classifier on Monte Carlo events. In this paper, we investigate the usage of Monte Carlo events that incorporate the dependence on the parameters of interest by reweighting. This enables more accurate likelihood learning with less training data and a more robust learning scheme that is more suited for automation and extensive deployment. We illustrate these advantages in the context of LHC precision probes of new Effective Field Theory interactions.

Neural collapse inspired semi-supervised learning with fixed classifier

Pseudo-labeling-based approaches are gaining prominence in Semi-Supervised Learning (SSL). Recent studies have identified that the key bottleneck in this methodology is addressing insufficient, incorrect, and imbalanced pseudo-labels. In this paper, we argue that the intrinsic problem behind this bottleneck is classifier bias, i.e., the classifier's prototypes suffer from poor uniformity. Further, inspired by neural collapse that reveals an optimal structure under supervised training scenarios, we address this classifier bias by utilizing an offline simplex Equiangular Tight Frame (ETF) classifier with maximally and equally separated prototypes. During the training phase, we maintain the prototypes of the classifier as fixed and concentrate on refining the feature encoder. Specifically, we integrate a straightforward clustering-based pseudo-labeling strategy with information maximization for feature learning. In practice, the fixed ETF classifier prevents the model from falling into a detrimental cycle, where a biased classifier exacerbates misaligned features, further perpetuating this bias. Furthermore, the clustering-based pseudo-labeling strategy reduces the dependency on complex threshold-adjusting mechanisms and effectively navigates the quantity-quality trade-off that plagues existing SSL methods. Leveraging these methodologies, we develop a simple yet powerful approach, termed ETF-SSL. Extensive experiments across Image, Text, and Audio datasets demonstrate that ETF-SSL can achieve competitive or superior performance compared to existing approaches. This success highlights the benefits of using a fixed ETF classifier in SSL and points to promising directions for future research in this area. The code is available at: https://github.com/yichenwang231/ETFSSL.

A Comprehensive Review of Tamaric gallica Linn.: A Herbal Haemostat with special reference in Unani Medicine

Background and aim: Tamarix gallica L. is commonly known as “Jhau”. These deciduous, herbaceous, twiggy shrub or small tree reaching up to about 5 meters high and mainly found in the salty regions and is found between interdunal areas of the desert. It has enormous potential and having a broad spectrum of activity on a variety of illnesses such as leukoderma, eye diseases, teeth disorder, any type of hemorrhage, spleen disorders etc. This review aims to compile maximum information available, regarding its various traditional uses, pharmacognosy, phytochemical constituents and pharmacological activities to enrich our knowledge about this plant.
 Materials and methods: A manual literature survey of classical Unani texts such as Khazā’in al-Adwiya, Tanqīh-al-Mufradāt, Bustān-al- Mufradāt, Muḥīṭ-i-A‘ẓam and online literature survey from Pubmed, Google scholar, Cochrane library, science direct was conducted to collect the information available on the plant Tamarix gallica. 
 Result and discussion: The findings of the literature survey on the basis of previous phytochemical analyses of this plant revealed the presence of a number of components, some of which are tannin, tannic acid, tamarixin, significant amount of phenolic and flavonoid content, antioxidants like carotenoids and essential oils which are extracted from various parts of the plants. Its reported pharmacological activities include anti-inflammatory, antinociceptive, anti-proliferative, antibacterial, antioxidant, anti-hyperlipidemic, antidiarrheal, cytotoxic activity and hepatoprotective activities.
 Keywords: Tamarix gallica, phytochemical analyses, tannin, antioxidant, pharmacological activities, anti-inflammatory, antinociceptive, antiproliferative, antibacterial.

Preserving shape details of pulse signals for video-based blood pressure estimation.

In recent years, imaging photoplethysmograph (iPPG) pulse signals have been widely used in the research of non-contact blood pressure (BP) estimation, in which BP estimation based on pulse features is the main research direction. Pulse features are directly related to the shape of pulse signals while iPPG pulse signals are easily disturbed during the extraction process. To mitigate the impact of pulse feature distortion on BP estimation, it is necessary to eliminate interference while retaining valuable shape details in the iPPG pulse signal. Contact photoplethysmograph (cPPG) pulse signals measured at rest can be considered as the undisturbed reference signal. Transforming the iPPG pulse signal to the corresponding cPPG pulse signal is a method to ensure the effectiveness of shape details. However, achieving the required shape accuracy through direct transformation from iPPG to the corresponding cPPG pulse signals is challenging. We propose a method to mitigate this challenge by replacing the reference signal with an average cardiac cycle (ACC) signal, which can approximately represent the shape information of all cardiac cycles in a short time. A neural network using multi-scale convolution and self-attention mechanisms is developed for this transformation. Our method demonstrates a significant improvement in the maximal information coefficient (MIC) between pulse features and BP values, indicating a stronger correlation. Moreover, pulse signals transformed by our method exhibit enhanced performance in BP estimation using different model types. Experiments are conducted on a real-world database with 491 subjects in the hospital, averaging 60 years of age.

Selection of HBV key reactivation factors based on maximum information coefficient combined with cosine similarity.

Hepatitis B Virus (HBV) reactivation is the most common complication for patients with primary liver cancer (PLC) after radiotherapy. How to reduce the reactivation of HBV has been a hot topic in the study of postoperative radiotherapy for liver cancer. To find out the inducement of HBV reactivation, a feature selection algorithm (MIC-CS) using maximum information coefficient (MIC) combined with cosine similarity (CS) was proposed to screen the risk factors that may affect HBV reactivation. Firstly, different factors were coded and MIC between patients was calculated to acquire the association between different factors and HBV reactivation. Secondly, a cosine similarity algorithm was constructed to calculate the similarity relationship between different factors, thus removing redundant information. Finally, combined with the weight of the two, the potential risk factors were sorted and the key factors leading to HBV reactivation were selected. The results indicated that HBV baseline, external boundary, TNM, KPS score, VD, AFP, and Child-Pugh could lead to HBV reactivation after radiotherapy. The classification model was constructed for the above factors, with the highest classification accuracy of 84% and the AUC value of 0.71. Comparing multiple feature selection methods, the results showed that the effect of the MIC-CS was significantly better than MIM, CMIM, and mRMR, so it has a very broad application prospect.

Few-Shot Learning for Crop Mapping from Satellite Image Time Series

Recently, deep learning methods have achieved promising crop mapping results. Yet, their classification performance is constrained by the scarcity of labeled samples. Therefore, the development of methods capable of exploiting label-rich environments to classify crops in label-scarce environments using only a few labeled samples per class is required. Few-shot learning (FSL) methods have achieved this goal in computer vision for natural images, but they remain largely unexplored in crop mapping from time series data. In order to address this gap, we adapted eight FSL methods to map infrequent crops cultivated in the selected study areas from France and a large diversity of crops from a complex agricultural area situated in Ghana. The FSL methods are commonly evaluated using class-balanced unlabeled sets from the target domain data (query sets), leading to overestimated classification results. This is unrealistic since these sets can have an arbitrary number of samples per class. In our work, we used the Dirichlet distribution to model the class proportions in few-shot query sets as random variables. We demonstrated that transductive information maximization based on α-divergence (α-TIM) performs better than the competing methods, including dynamic time warping (DTW), which is commonly used to tackle the lack of labeled samples. α-TIM achieved, for example, a macro F1-score of 59.6% in Ghana in a 24-way 20-shot setting (i.e., 20 labeled samples from each of the 24 crop types) and a macro F1-score of 75.9% in a seven-way 20-shot setting in France, outperforming the second best-performing methods by 2.7% and 5.7%, respectively. Moreover, α-TIM outperformed a baseline deep learning model, highlighting the benefits of effectively integrating the query sets into the learning process.

Urban Water Demand Prediction Based on Attention Mechanism Graph Convolutional Network-Long Short-Term Memory

Predicting short-term urban water demand is essential for water resource management and directly impacts urban water resource planning and supply–demand balance. As numerous factors impact the prediction of short-term urban water demand and present complex nonlinear dynamic characteristics, the current water demand prediction methods mainly focus on the time dimension characteristics of the variables, while ignoring the potential influence of spatial characteristics on the temporal characteristics of the variables. This leads to low prediction accuracy. To address this problem, a short-term urban water demand prediction model which integrates both spatial and temporal characteristics is proposed in this paper. Firstly, anomaly detection and correction are conducted using the Prophet model. Secondly, the maximum information coefficient (MIC) is used to construct an adjacency matrix among variables, which is combined with a graph convolutional neural network (GCN) to extract spatial characteristics among variables, while a multi-head attention mechanism is applied to enhance key features related to water use data, reducing the influence of unnecessary factors. Finally, the prediction of short-term urban water demand is made through a three-layer long short-term memory (LSTM) network. Compared with existing prediction models, the hybrid model proposed in this study reduces the average absolute percentage error by 1.868–2.718%, showing better prediction accuracy and prediction effectiveness. This study can assist cities in rationally allocating water resources and lay a foundation for future research.

The relative and combined ability of stress hyperglycemia ratio and N-terminal pro-B-type natriuretic peptide to predict all-cause mortality in diabetic patients with multivessel coronary artery disease

BackgroundStress hyperglycemia ratio (SHR) and N-terminal pro-B-type natriuretic peptide (NT-proBNP) are independently associated with increased mortality risk in diabetic patients with coronary artery disease (CAD). However, the role of these biomarkers in patients with diabetes and multivessel disease (MVD) remains unknown. The present study aimed to assess the relative and combined abilities of these biomarkers to predict all-cause mortality in patients with diabetes and MVD.MethodsThis study included 1148 diabetic patients with MVD who underwent coronary angiography at Tianjin Chest Hospital between January 2016 and December 2016. The patients were divided into four groups according to their SHR (SHR-L and SHR-H) and NT-proBNP (NT-proBNP-L and NT-proBNP-H) levels. The primary outcome was all-cause mortality. Multivariate Cox regression analyses were performed to evaluate the association of SHR and NT-proBNP levels with all-cause mortality.ResultsDuring a mean 4.2 year follow-up, 138 patients died. Multivariate analysis showed that SHR and NT-proBNP were strong independent predictors of all-cause mortality in diabetic patients with MVD (SHR: HR hazard ratio [2.171; 95%CI 1.566–3.008; P < 0.001; NT-proBNP: HR: 1.005; 95%CI 1.001–1.009; P = 0.009). Compared to patients in the first (SHR-L and NT-proBNP-L) group, patients in the fourth (SHR-H and NT-proBNP-H) group had the highest mortality risk (HR: 12.244; 95%CI 5.828–25.721; P < 0.001). The areas under the curve were 0.615(SHR) and 0.699(NT-proBNP) for all-cause mortality. Adding either marker to the original models significantly improved the C-statistic and integrated discrimination improvement values (all P < 0.05). Moreover, combining SHR and NT-proBNP levels into the original model provided maximal prognostic information.ConclusionsSHR and NT-proBNP independently and jointly predicted all-cause mortality in diabetic patients with MVD, suggesting that strategies to improve risk stratification in these patients should incorporate SHR and NT-porBNP into risk algorithms.

Expressway lane change strategy of autonomous driving based on prior knowledge and data-driven

Automatic driving in expressways is generally considered to be the easiest for commercial landing, and vehicle behavior decision-making is the core of automatic driving technology, which directly affects the safety and comfort of vehicle driving. In this paper, an automatic lane change strategy based on prior knowledge and data-driven is proposed for expressway. Our method decouples autonomous driving decisions into two processes. Firstly, we propose an optimization model of lane change features based on prior knowledge, which not only provides a basis for the selection of vehicle features but also effectively reduces the interference of invalid features. The driving risk field is used to screen the vehicle targets that have potential influence on the lane change of the ego vehicle, and the maximal information coefficient evaluates the effectiveness of the features of the vehicle targets through the maximal information coefficient to optimize the lane change features. Then, a data-driven lane-changing model is proposed based on Attention-BiLSTM (Bi-directional Long Short-Term Memory), which transforms vehicle lane-changing into time series prediction, fully explores the relationship between lane-changing features in the context information, and effectively improves the anti-interference ability of the system against accidental errors. In addition, the attention mechanism can adaptively adjust the weight of lane-changing features and effectively capture the important information that needs attention in different lane-changing states, further improving the accuracy of the model. Finally, the validity of the model is verified by remote sensing data sets and real vehicle experiments. The test results show that the test accuracy of our method is 95.1% in data sets and 94.2% in real vehicle experiments, which fully meets the requirements of lane change decisions about automatic driving.

Noise aware content-noise complementary GAN with local and global discrimination for low-dose CT denoising

In response to rising concerns over radiation exposure in computed tomography (CT) imaging, effective denoising methods for low-dose CT (LDCT) images are crucial. In recent years, the use of deep learning techniques especially generative adversarial networks (GANs) significantly enhanced the efficiency of LDCT denoising methods, surpassing traditional methods. However, GAN-based denoising methods often face challenges in preserving structural consistency and fine details. This study introduces a novel GAN framework with three accretions to enhance the effectiveness of LDCT denoising. Firstly, our generator is designed to leverage a complementary learning scheme between image noise and image content via two distinct paths. One path focuses on exploring the anatomical information of the image, while the second path is dedicated to learning the noise pattern. This complementary learning scheme provides stable noise cancellation while preserving the maximum structural information of the image. Subsequently, we propose a novel noise-conscious mean absolute error loss to address the challenge posed by the non-stationary characteristic of CT noise. In contrast to conventional MAE loss, this loss attentively prioritizes the different parts of the image based on the local distribution of noise in that region. We also incorporate a gradient-domain loss into the loss function, which inspires the generator to preserve precise image details through explicit guidance. Finally, this study adopted a U-Net-based design for the discriminator to better regularize the model by discriminating between the clean image and the denoised image at both global and local levels. The merit of this discriminator is that it can better adapt to the non-stationary environment of GAN training and guide the generator to produce denoised images that are locally and globally consistent. Our thorough experiments using abdominal CT and lung CT datasets demonstrate the superior performance of our approach compared to state-of-the-art methods.

IRT for voting advice applications: a multi-dimensional test that is adaptive and interpretable

Voting advice applications rely on user input to match user preferences to political parties or candidates. Providing the input can be time-consuming, which may have a negative effect on participation. For individuals who are under time constraints or who are affected by survey fatigue, the participation threshold may be lowered if there is an option to conclude the test without answering all question items. The test result should ideally be close to the result that the participant would have gotten after answering the full battery of questions. We propose a method that allows respondents to conclude a VAA early and still get results with sufficient accuracy.The method proposed here extends the Graded Response Model and the Maximum Information Criterion, used in Item Response Theory. The aim of the method is to allow the user to control the length of the test. Furthermore, we want a simpler interpretation of multidimensional parameter estimates than we get from traditional MIRT. To achieve this, we propose an algorithm for adaptive IRT capable of selecting from a pool of items that belong to separate unidimensional scales. Using both simulated data and response data from a voting advice application project, we evaluate the accuracy of shorter tests implemented with our adaptive method. When only a few test items are answered, our proposed method outperforms a static-order IRT test of equal length in identifying the best match. We expect that implementation of the method can increase participation and engagement in voting advice applications.

A Feature Selection Method Based on Graph Theory for Cancer Classification.

Gene expression profile data is a good data source for people to study tumors, but gene expression data has the characteristics of high dimension and redundancy. Therefore, gene selection is a very important step in microarray data classification. In this paper, a feature selection method based on the maximum mutual information coefficient and graph theory is proposed. Each feature of gene expression data is treated as a vertex of the graph, and the maximum mutual information coefficient between genes is used to measure the relationship between the vertices to construct an undirected graph, and then the core and coritivity theory is used to determine the feature subset of gene data. In this work, we used three different classification models and three different evaluation metrics such as accuracy, F1-Score, and AUC to evaluate the classification performance to avoid reliance on any one classifier or evaluation metric. The experimental results on six different types of genetic data show that our proposed algorithm has high accuracy and robustness compared to other advanced feature selection methods. In this method, the importance and correlation of features are considered at the same time, and the problem of gene selection in microarray data classification is solved.

Script late injection: a framework to introduce JavaScript into web pages

Script injection is one type of fault present in web, which mostly utilizes user data to execute code without applying any type of filters. Script injection can impact both client and server making exposing them to vulnerabilities. Security and related products may need to execute logic on the client-side generally in a browser. In order to achieve this, proxy servers inject appropriate JavaScript code into the responses they proxy. Typically, the injection point is at the end of the body element. The framework introduced in this paper rather uses a stack-based approach to determine the injection point in the web page. Ten kilobytes from the end of a web page are given as a string input to the framework, after tokenization and construction of the vector of tokens. A stack is used to determine the injection point. Along with the position of the injection point, a warning flag is also estimated indicating the correctness of the injection point. Different types of web pages were considered for running the unit tests and fuzzy tests on the framework. These classes of pages are determined by crawling most used web pages. The injected scripts are executed once the body content is completely loaded. Hence, it can retrieve maximum information without affecting end-user performance. It also does the job at a low cost.

Multicamera Collaboration for 3-D Visualization via Correlated Information Maximization

Multicamera Collaboration for 3-D Visualization via Correlated Information Maximization

Skin lesion classification based on hybrid self‐supervised pretext task

AbstractThe combination of observation of skin lesion and digital image technology contributes to the diagnosis and treatment of skin diseases. To solve the problems of large variation of target size and shape in skin disease images, small difference between disease images and normal images, and difficulty of label acquisition, we propose a classification algorithm for skin lesion based on hybrid self‐supervised pretext tasks. We combine the discriminative self‐supervised relational reasoning task with the generative self‐supervised mutual information maximization task and design the loss function. Experiments show that the generative‐discriminative self‐supervised co‐training algorithm can identify images with small differences and learn discriminative features and achieve a classification accuracy of 82.6% on the DermaMNIST dataset.

Anthropology of mental illnesses: the origin of psychopathologies and developmental psychopathology: gender issues

This research aims to address the potential factors that generate mental disorders. Human pathologies have been extensively studied and researched from a biological perspective, however many scientific studies in the field of developmental psychopathology indicate that political, religious and biopsychosocial-cultural factors can influence both the origin and the course of evolution of mental illnesses. The methodology used was a bibliographic review in the field of psychology, psychiatry, psychoanalysis and anthropology. Interdisciplinary and multidisciplinary studies that integrate disciplines such as anthropology, psychology, psychiatry, psychoanalysis, education, social assistance and sociology are fundamental not only for nosology and nosography, but above all with the support of maximum information on how factors act genetic, hormonal and biological factors and cultural and socioeconomic agents interact in the generation of mental disorders at the individual and collective level, taking precedence over the effects of social disorders in the generation of psychopathologies in the LGBTQIAPN+ population.