Increased Information Content Research Articles

Simplified Information Maximization for Improving Generalization Performance in Multilayered Neural Networks

A new type of information-theoretic method is proposed to improve prediction performance in supervised learning. The method has two main technical features. First, the complicated procedures used to increase information content are replaced by the direct use of hidden neuron outputs. Information is controlled by directly changing the outputs of the hidden neurons. In addition, to simultaneously increase information content and decrease errors between targets and outputs, the information acquisition and use phases are separated. In the information acquisition phase, the autoencoder tries to acquire as much information content on input patterns as possible. In the information use phase, information obtained in the acquisition phase is used to train supervised learning. The method is a simplified version of actual information maximization and directly deals with the outputs from neurons. The method was applied to the three data sets, namely, Iris, bankruptcy, and rebel participation data sets. Experimental results showed that the proposed simplified information acquisition method was effective in increasing the real information content. In addition, by using the information content, generalization performance was greatly improved.

Fractal Analysis, Information-Theoretic Similarities and SVM Classification for Multichannel, Multi-Frequency Pre-Seismic Electromagnetic Measurements

A multichannel, multi-frequency approach on the analysis of critical electromagnetic (EM) signatures prior to earthquakes is presented. The algorithm employed is based on single-channel techniques for the identification of longmemory trends in fractal processes and attempts to take advantage of the increased information content that is provided by multichannel EM recordings. The EM measurements consist of four channels which correspond to four distinct EM radiation frequencies. Two of these frequencies lie in the kHz range and the other two in the MHz range. Our analysis of a three-month EM activity period shows that there exists some degree of similarity between EM channels that are close in frequency, in terms of an information theoretic measure. More importantly, the multichannel-based detection results seem to be in close agreement with the main earthquake occurrences of the three-month period. The combined output of the multiple channels is used to train a Support Vectors Machine (SVM) classifier in order to identify precursory EM signal segments of forthcoming seismic events and a high accuracy rate is reported.

Glycan heterogeneity on gold nanoparticles increases lectin discrimination capacity in label-free multiplexed bioassays.

The development of new analytical tools as point-of-care biosensors is crucial to combat the spread of infectious diseases, especially in the context of drug-resistant organisms, or to detect biological warfare agents. Glycan/lectin interactions drive a wide range of recognition and signal transduction processes within nature and are often the first site of adhesion/recognition during infection making them appealing targets for biosensors. Glycosylated gold nanoparticles have been developed that change colour from red to blue upon interaction with carbohydrate-binding proteins and may find use as biosensors, but are limited by the inherent promiscuity of some of these interactions. Here we mimic the natural heterogeneity of cell-surface glycans by displaying mixed monolayers of glycans on the surface of gold nanoparticles. These are then used in a multiplexed, label-free bioassay to create ‘barcodes’ which describe the lectin based on its binding profile. The increased information content encoded by using complex mixtures of a few sugars, rather than increased numbers of different sugars makes this approach both scalable and accessible. These nanoparticles show increased lectin identification power at a range of lectin concentrations, relative to single-channel sensors. It was also found that some information about the concentration of the lectins can be extracted, all from just a simple colour change, taking this technology closer to being a realistic biosensor.

Optimal Design for Reactivity Ratio Estimation: A Comparison of Techniques for AMPS/Acrylamide and AMPS/Acrylic Acid Copolymerizations

Water-soluble polymers of acrylamide (AAm) and acrylic acid (AAc) have significant potential in enhanced oil recovery, as well as in other specialty applications. To improve the shear strength of the polymer, a third comonomer, 2-acrylamido-2-methylpropane sulfonic acid (AMPS), can be added to the pre-polymerization mixture. Copolymerization kinetics of AAm/AAc are well studied, but little is known about the other comonomer pairs (AMPS/AAm and AMPS/AAc). Hence, reactivity ratios for AMPS/AAm and AMPS/AAc copolymerization must be established first. A key aspect in the estimation of reliable reactivity ratios is design of experiments, which minimizes the number of experiments and provides increased information content (resulting in more precise parameter estimates). However, design of experiments is hardly ever used during copolymerization parameter estimation schemes. In the current work, copolymerization experiments for both AMPS/AAm and AMPS/AAc are designed using two optimal techniques (Tidwell-Mortimer and the error-in-variables-model (EVM)). From these optimally designed experiments, accurate reactivity ratio estimates are determined for AMPS/AAm (rAMPS = 0.18, rAAm = 0.85) and AMPS/AAc (rAMPS = 0.19, rAAc = 0.86).

Light-assisted small-molecule screening against protein kinases.

High-throughput live-cell screens are intricate elements of systems biology studies and drug discovery pipelines. Here, we demonstrate an optogenetics-assisted method that obviates the addition of chemical activators and reporters, reduces the number of operational steps and increases information content in a cell-based small molecule screen against human protein kinases including an orphan receptor tyrosine kinase. This blueprint for all-optical screening can be adapted to many drug targets and cellular processes.

Photopolarimetric retrievals of snow properties

Abstract. Polarimetric observations of snow surfaces, obtained in the 410–2264 nm range with the Research Scanning Polarimeter onboard the NASA ER-2 high-altitude aircraft, are analyzed and presented. These novel measurements are of interest to the remote sensing community because the overwhelming brightness of snow plagues aerosol and cloud retrievals based on airborne and spaceborne total reflection measurements. The spectral signatures of the polarized reflectance of snow are therefore worthwhile investigating in order to provide guidance for the adaptation of algorithms currently employed for the retrieval of aerosol properties over soil and vegetated surfaces. At the same time, the increased information content of polarimetric measurements allows for a meaningful characterization of the snow medium. In our case, the grains are modeled as hexagonal prisms of variable aspect ratios and microscale roughness, yielding retrievals of the grains' scattering asymmetry parameter, shape and size. The results agree with our previous findings based on a more limited data set, with the majority of retrievals leading to moderately rough crystals of extreme aspect ratios, for each scene corresponding to a single value of the asymmetry parameter.

Stable isotope labeling strategy based on coding theory.

We describe a strategy for stable isotope-aided protein nuclear magnetic resonance (NMR) analysis, called stable isotope encoding. The basic idea of this strategy is that amino-acid selective labeling can be considered as "encoding and decoding" processes, in which the information of amino acid type is encoded by the stable isotope labeling ratio of the corresponding residue and it is decoded by analyzing NMR spectra. According to the idea, the strategy can diminish the required number of labelled samples by increasing information content per sample, enabling discrimination of 19 kinds of non-proline amino acids with only three labeled samples. The idea also enables this strategy to combine with information technologies, such as error detection by check digit, to improve the robustness of analyses with low quality data. Stable isotope encoding will facilitate NMR analyses of proteins under non-ideal conditions, such as those in large complex systems, with low-solubility, and in living cells.

ACCOUNTING FUNDAMENTALS AND VARIATIONS OF STOCK PRICE: FORWARD LOOKING INFORMATION INDUCEMENT

This study investigates a permanent issue about low association between accounting fundamentals and variations of stock prices. It induces not only historical accounting fundamentals, but also forward looking information. Investors consider forward looking information that enables them to predict potential future cash flow, increase predictive power, lessen mispricing error, increase information content and drives future price equilibrium. The accounting fundamentals are earnings yield, book value, profitability, growth opportunities and discount rate or they could be called as five-related-cash flow factors. The forward looking information are expected earnings and expected growth opportunities. This study suggests that model inducing forward looking information could improve association degree between accounting fundamentals and the movements of stock prices. In other words, they have higher value relevance than not by inducing. Finally, this study concludes that inducing forward looking information could predict stock price accurately and reduce stock price deviations from their fundamental value. It also implies that trading strategies should realize to firm’s future rational expectations. Keywords: earnings yield, book value, profitability, growth opportunities, discount rate, accounting fundamentals, forward looking, value relevance

An integrated approach to reconstructing genome-scale transcriptional regulatory networks.

Transcriptional regulatory networks (TRNs) program cells to dynamically alter their gene expression in response to changing internal or environmental conditions. In this study, we develop a novel workflow for generating large-scale TRN models that integrates comparative genomics data, global gene expression analyses, and intrinsic properties of transcription factors (TFs). An assessment of this workflow using benchmark datasets for the well-studied γ-proteobacterium Escherichia coli showed that it outperforms expression-based inference approaches, having a significantly larger area under the precision-recall curve. Further analysis indicated that this integrated workflow captures different aspects of the E. coli TRN than expression-based approaches, potentially making them highly complementary. We leveraged this new workflow and observations to build a large-scale TRN model for the α-Proteobacterium Rhodobacter sphaeroides that comprises 120 gene clusters, 1211 genes (including 93 TFs), 1858 predicted protein-DNA interactions and 76 DNA binding motifs. We found that ~67% of the predicted gene clusters in this TRN are enriched for functions ranging from photosynthesis or central carbon metabolism to environmental stress responses. We also found that members of many of the predicted gene clusters were consistent with prior knowledge in R. sphaeroides and/or other bacteria. Experimental validation of predictions from this R. sphaeroides TRN model showed that high precision and recall was also obtained for TFs involved in photosynthesis (PpsR), carbon metabolism (RSP_0489) and iron homeostasis (RSP_3341). In addition, this integrative approach enabled generation of TRNs with increased information content relative to R. sphaeroides TRN models built via other approaches. We also show how this approach can be used to simultaneously produce TRN models for each related organism used in the comparative genomics analysis. Our results highlight the advantages of integrating comparative genomics of closely related organisms with gene expression data to assemble large-scale TRN models with high-quality predictions.

Adaptive Decomposition of Highly Resolved Time Series into Local and Non-Local Components

ABSTRACTHighly time-resolved air monitoring data are widely being collected over long time horizons in order to characterize ambient and near-source air quality trends. In many applications, it is desirable to split the time-resolved data into two or more components (e.g., local and regional) for apportionment and mitigation purposes. While there may be increased information content in highly time-resolved data, the temporal resolution may also increase entropic effects on the data, thereby dramatically clouding the very information sought in time-resolved data. Specialized methods such as filtering may be required to extract the underlying information content. Constrained and Adaptive Decomposition of Time Series (CADETS) is a new method that can help carve out components of time series based on the content of the frequencies present in the time series. CADETS is also a flexible approach that allows the user to choose the bifurcation point with minimal negative impacts. Using this algorithm, we demonstrate that a time series signal may be decomposed into two useful and interpretable signals that can help identify aspects that may otherwise be hidden or distorted. Using the output from the CADETS algorithm, we show that ultrafine particles (30–100 nm) collected near a major highway may be split into a 64:36 ratio of highly varying (local) and slowly varying (regional) components, meanwhile identical measurements at a background location were estimated to split into a 56:44 local versus regional ratio.

Bayesian Design of Experiments Applied to a Complex Polymerization System: Nitrile Butadiene Rubber Production in a Train of CSTRs

Bayesian design of experiments can be very useful for complex polymerizations and other chemical engineering processes. The technique has many practical benefits; it incorporates prior information, allows for adjustment of design levels, increases the information content, and optimizes experimental resources. In this work, Bayesian design is applied to the simulated emulsion copolymerization of NBR in a series of CSTRs. Statistical comparisons show that the Bayesian design is as good as (or better than) standard design techniques. This makes the Bayesian design superior overall, as it provides the extra flexibility of designing sequences of fewer trials and an increased information content.

Multiplexed microfluidic enzyme assays for simultaneous detection of lipolysis products from adipocytes.

Microfluidics has enabled new cell biology experiments. Incorporating chemical monitoring of cellular secretion into chips offers the potential to increase information content and utility of such systems. In this work, an integrated, multilayer polydimethylsiloxane microfluidic chip was developed to simultaneously measure fatty acids and glycerol secreted from cultured adipocytes on chip in near real time. Approximately 48,000 adipocytes were loaded into a cell chamber in a reversibly sealed chip. Cells were perfused at 0.75 μL/min. Cell perfusate was split and directed to separate, continuously operating fluorescent enzyme assay channel networks. The fluorescent assay products were detected simultaneously near the outlet of the chip. The fatty acid and glycerol assays had linear dynamic ranges of 150 and 110 μM and limit of detection (LOD) of 6 and 5 μM, respectively. Surface modifications including pretreatment with sodium dodecyl sulfate were utilized to prevent adsorption of fatty acids to the chip surface. Using the chip, basal fatty acid and glycerol concentrations ranged from 0.18 to 0.7 nmol × 10(6) cell(-1) min(-1) and from 0.23 to 0.85 nmol × 10(6) cell(-1) min(-1), respectively. Using valves built into the chip, the perfusion solution was switched to add 20 μM isoproterenol, a β-adrenergic agonist, which stimulates the release of glycerol and fatty acids in adipocytes. This manipulation resulted in a rapid and stable 1.5- to 6.0-fold increase of non-esterified fatty acid (NEFA) and glycerol. The ratio of NEFA:glycerol released increased with adipocyte age. These experiments illustrate the potential for performing multiple real-time assays on cells in culture using microfluidic devices.

The competing benefits of noise and heterogeneity in neural coding.

Noise and heterogeneity are both known to benefit neural coding. Stochastic resonance describes how noise, in the form of random fluctuations in a neuron's membrane voltage, can improve neural representations of an input signal. Neuronal heterogeneity refers to variation in any one of a number of neuron parameters and is also known to increase the information content of a population. We explore the interaction between noise and heterogeneity and find that their benefits to neural coding are not independent. Specifically, a neuronal population better represents an input signal when either noise or heterogeneity is added, but adding both does not always improve representation further. To explain this phenomenon, we propose that noise and heterogeneity operate using two shared mechanisms: (1) temporally desynchronizing the firing of neurons in the population and (2) linearizing the response of a population to a stimulus. We first characterize the effects of noise and heterogeneity on the information content of populations of either leaky integrate-and-fire or FitzHugh-Nagumo neurons. We then examine how the mechanisms of desynchronization and linearization produce these effects, and find that they work to distribute information equally across all neurons in the population in terms of both signal timing (desynchronization) and signal amplitude (linearization). Without noise or heterogeneity, all neurons encode the same aspects of the input signal; adding noise or heterogeneity allows neurons to encode complementary aspects of the input signal, thereby increasing information content. The simulations detailed in this letter highlight the importance of heterogeneity and noise in population coding, demonstrate their complex interactions in terms of the information content of neurons, and explain these effects in terms of underlying mechanisms.

Modification of filters of multispectral raster images to solve the problem of remote probing

Methods of remote Еarth probing (REP) is widely used to solve various problems. There are areas of application of remote sensing: the search of minerals, information on the environment, assessing the state, the study of ecosystems and their diversity, hydrography, evaluation of consequences of natural disasters. There are more than 20 companies in the world involved in REP tasks, including: Jeppesen, Trimble, ArcGis, GeoGude, Spectra Precision.Modern remote Earth probing devices from high-flying carriers capture images in the visible and infrared ranges. Altogether brightness distribution of corresponding images form multidimensional geometric object – multispectral image (MSI), the components of which give the spatial form and physical condition of recorded material objects.The main method of reducing the dimension MSI presented in this format, currently are the filtering of multiple components, which leads to the approximation of the brightness distribution of the source bitmap set of graphics primitives.The task of increasing of remote probing image information content in each case requires individual approach to images using the combined information processing algorithms, since there is no universal approach that would be able to provide a high level descramble condition of such images.To solve the problem of remote probing, namely providing high information content and precision there is lack of universal tool (filtration method) that would be able to provide a high level of ability to search the selected unit (airport, car, coastline) regardless of the angle, the picture quality of the original image and computing power. Currently almost all the problems related to information content and accuracy is a heuristic task that requires an individual approach to each shot.Studies on the example of airport ESSA (Stockholm) provided by Jeppesen (USA) showed that after addition of local filtering methods using selective smoothing, improved signal/noise ratio (decreased effect of analog noise) and blur the contours of the object has almost no effect on the information content, which increased the distinctive ability and made it possible to clearly identify the object (runway).Analysis and modification of existing filters and correction of existing search methods for increasing information content of multispectral raster images (search of small object) based on filtering polychromatic shades and brightness thresholds.The use of modified filters and search methods made it possible to create a software product, which greatly improved the images descramblecondition, making it possible to differentiate small objects that cannot be seen on the original images, and select objects with thermal anomalies and objects which slightly differ in color in the visible range on the original images.

Developing a consistent process‐based conceptualization of catchment functioning using measurements of internal state variables

AbstractWe use isotope data in addition to discharge and groundwater level data to conceptualize the internal processes of runoff generation and tracer transport in a low parameter coupled flow‐tracer model that could predict the runoff response and isotopic composition of an upland stream. We used sensitivity analysis to assess the effect of these data on model calibration in terms of parameter identifiability and the model's ability to predict the stream's runoff response, isotopic composition and water age. The results showed that the incorporation of tracer data in particular, clearly increased parameter identifiability and improved the predictive power of models for simulating both streamflow and isotopes. This also resulted in a more consistent process‐based conceptualization of catchment functioning. We could also show that using models as learning tools can guide sampling campaigns toward measurements with increased information content for further modeling. We conclude that this is a promising approach for assessing dominant processes in coupled flow‐tracer models. This is of value when such models are being used to test hypotheses about the hydrological functioning of catchments, particularly in relation to pollutant transfers.

Optical fluctuation microscopy based on calculating local entropy values

We demonstrate a novel and easy-to-use method to dramatically reduce noise and background contributions in advanced fluorescence microscopy experiments. The underlying idea is that the entropy value increases for systems with a large number of accessible energy states. Intensity fluctuations originating from photophysical or photochemical effects lead to an increased information content. Calculating the pixel-wise entropy value results in an enhancement of the signal-to-noise ratio by a factor of 90–100. Comparing ECI (entropy-based contrast-enhanced imaging) to superresolution methods such as STORM and SOFI, we find that this technique also bears substantial potential for enhancing fluctuation-based superresolution microscopies.

Variability and information content in auditory cortex spike trains during an interval-discrimination task

Processing of temporal information is key in auditory processing. In this study, we recorded single-unit activity from rat auditory cortex while they performed an interval-discrimination task. The animals had to decide whether two auditory stimuli were separated by either 150 or 300 ms and nose-poke to the left or to the right accordingly. The spike firing of single neurons in the auditory cortex was then compared in engaged vs. idle brain states. We found that spike firing variability measured with the Fano factor was markedly reduced, not only during stimulation, but also in between stimuli in engaged trials. We next explored if this decrease in variability was associated with an increased information encoding. Our information theory analysis revealed increased information content in auditory responses during engagement compared with idle states, in particular in the responses to task-relevant stimuli. Altogether, we demonstrate that task-engagement significantly modulates coding properties of auditory cortical neurons during an interval-discrimination task.

Multiobjective Framework for Model-based Design of Experiments to Improve Parameter Precision and Minimize Parameter Correlation

The need for first principles based models for chemical and biological processes has led to the development of techniques for model-based design of experiments (MBDOE). These techniques help in speeding up the parameter estimation efforts and typically lead to improved parameter precision with a relatively short experimental campaign. In the case of complex kinetic networks involving parallel and/or consecutive reactions, correlation among model parameters makes the inverse problem of parameter estimation very difficult. It is therefore important to develop experimental design techniques that not only increase information content about the system to facilitate precise parameter estimation but also reduce the correlation among parameters. This article presents a multiobjective optimization (MOO) based framework for experimental design, where, in addition to the traditional objective of eliciting maximally informative data for parameter estimation, an explicit objective to reduce correlation among parameter...

Testing the predictive performance of distribution models

Distribution models are used to predict the likelihood of occurrence or abundance of a species at locations where census data are not available. An integral part of modelling is the testing of model performance. We compared different schemes and measures for testing model performance using 79 species from the North American Breeding Bird Survey. The four testing schemes we compared featured increasing independence between test and training data: resubstitution, random data hold‐out and two spatially segregated data hold‐out designs. The different testing measures also addressed different levels of information content in the dependent variable: regression R2 for absolute abundance, squared correlation coefficient r2 for relative abundance and AUC/Somer’s D for presence/absence. We found that higher levels of independence between test and training data lead to lower assessments of prediction accuracy. Even for data collected independently, spatial autocorrelation leads to dependence between random hold‐out test data and training data, and thus to inflated measures of model performance. While there is a general awareness of the importance of autocorrelation to model building and hypothesis testing, its consequences via violation of independence between training and testing data have not been addressed systematically and comprehensively before. Furthermore, increasing information content (from correctly classifying presence/absence, to predicting relative abundance, to predicting absolute abundance) leads to decreasing predictive performance. The current tests for presence/absence distribution models are typically overly optimistic because a) the test and training data are not independent and b) the correct classification of presence/absence has a relatively low information content and thus capability to address ecological and conservation questions compared to a prediction of abundance. Meaningful evaluation of model performance requires testing on spatially independent data, if the intended application of the model is to predict into new geographic or climatic space, which arguably is the case for most applications of distribution models.

Simultaneous Detection, Genotyping, and Quantification of Human Papillomaviruses by Multicolor Real-Time PCR and Melting Curve Analysis

Long-term infection with high-risk human papillomavirus (HPV) is the leading cause of cervical cancer, while infection with low-risk HPV is the major reason for condylomata acuminata. An accurate, rapid, and convenient assay that is able to simultaneously detect, genotype, and quantify HPV would be of great clinical value yet remains to be achieved. We developed a three-color real-time PCR assay that is able to analyze 30 predominant HPV types in three reactions. The amplification curves indicated the presence of HPV, melting curve analysis identified the HPV genotype, and the quantification cycle value determined the quantity. We applied this assay to 647 cervical swab samples, and the results were compared with those obtained with a commercial genotyping system. The proposed assay had a limit of detection of 5 to 50 copies per reaction and a dynamic range of 5 × 10(1) to 5 × 10(6) copies per reaction. A comparison study showed that the overall sample concordance with the comparison method was 91.6% and the type agreement was greater than 98.7%. The quantification study demonstrated that the loads of HPV type 16 in 30 samples with cervical intraepithelial neoplasia grade III (CIN III) lesions were significantly higher than those in samples with CIN I lesions or CIN II lesions, and the results were concordant with those of the comparison method. The increased information content, high throughput, and low cost would facilitate the use of this real-time PCR-based assay in a variety of clinical settings.