Binary Units Research Articles

Hyperparameter tuning for hidden unit conditional random fields

PurposeIn this paper, the problem of a nonlinear model – specifically the hidden unit conditional random fields (HUCRFs) model, which has binary stochastic hidden units between the data and the labels – exhibiting unstable performance depending on the hyperparameter under consideration.Design/methodology/approachThere are three main optimization search methods for hyperparameter tuning: manual search, grid search and random search. This study shows that HUCRFs’ unstable performance depends on the hyperparameter values used and its performance is based on tuning that draws on grid and random searches. All experiments conducted used the n-gram features – specifically, unigram, bigram, and trigram.FindingsNaturally, selecting a list of hyperparameter values based on a researchers’ experience to find a set in which the best performance is exhibited is better than finding it from a probability distribution. Realistically, however, it is impossible to calculate using the parameters in all combinations. The present research indicates that the random search method has a better performance compared with the grid search method while requiring shorter computation time and a reduced cost.Originality/valueIn this paper, the issues affecting the performance of HUCRF, a nonlinear model with performance that varies depending on the hyperparameters, but performs better than CRF, has been examined.

Region-based facial representation for real-time Action Units intensity detection across datasets

Most research on facial expressions recognition has focused on binary Action Units (AUs) detection, while graded changes in their intensity have rarely been considered. This paper proposes a method for the real-time detection of AUs intensity in terms of the Facial Action Coding System scale. It is grounded on a novel and robust anatomically based facial representation strategy, for which features are registered from a different region of interest depending on the AU considered. Real-time processing is achieved by combining Histogram of Gradients descriptors with linear kernel Support Vector Machines. Following this method, AU intensity detection models are built and validated through the DISFA database, outperforming previous approaches without real-time capabilities. An in-depth evaluation through three different databases (DISFA, BP4D and UNBC Shoulder-Pain) further demonstrates that the proposed method generalizes well across datasets. This study also brings insights about existing public corpora and their impact on AU intensity prediction.

The Binary Energy Harvesting Channel With a Unit-Sized Battery

We consider a binary energy harvesting communication channel with a finite-sized battery at the transmitter. In this model, the channel input is constrained by the available energy at each channel use, which is driven by an external energy harvesting process, the size of the battery, and the previous channel inputs. We consider an abstraction where energy is harvested in binary units and stored in a battery with the capacity of a single unit, and the channel inputs are binary. Viewing the available energy in the battery as a state, this is a state-dependent channel with input-dependent states, memory in the states, and causal state information available at the transmitter only. We find an equivalent representation for this channel based on the timings of the symbols, and determine the capacity of the resulting equivalent timing channel via an auxiliary random variable. We present achievable rates based on certain selections of this auxiliary random variable, which resemble lattice coding for the timing channel. We develop upper bounds for the capacity by using a genie-aided method, and also by quantifying the leakage of the state information to the receiver. We show that the proposed achievable rates are asymptotically capacity achieving for small energy harvesting rates. We extend the results to the case of ternary channel inputs. We numerically observe that our achievable rates are notably close to the upper bounds, and outperform basic Shannon strategies that only consider instantaneous battery states, for all parameter values.

A method for localizing a reference object in a current image with several bright objects

To ensure the effective functioning of the correlation-extreme navigation systems (CENSs), a method is developed to localize a reference object (RO) in a current image (CI) with several bright objects. The peculiarity of the method consists in converting the CI to a binary unit by determining the average value of the background and setting it for the threshold of the image quantization, which in turn determines the amount of probabilities of errors of the first and second kinds as well as entails assigning the objects of the viewing surface (VS) and the backgrounds to two classes: the RO and the background. The CI model is represented by the brightness values of the corresponding objects and backgrounds of the VS in the differentiation elements. In the model of the current image, the RO has the highest brightness. Other objects that are similar in brightness and commensurate with the RO are categorized as false. The reference image (RI) is set by the contrast mark and the geometrical shape of the object, and it is binary. An algorism has been developed for localizing the RO in an image by searching for a fragment of a binary CI with a maximum value of units that coincides with the RI. The peculiarity of the algorithm consists in adapting the application procedure for the threshold conversion of a CI with an unknown value of the signal-noise ratio. A method has been developed to clarify the maximum value of the DF and to determine the coordinates of the RO in the field of the CI matrix. The method consists in the summation of the number of units of different sections and finding the highest value of the DF. The highest value of the DF coincides with the full match between the CI and the RI. An analytical expression has been obtained for the estimation of the probability of localizing the RO. The expression establishes dependence of the probability of localizing the RO on the parameters that are specified in the stages of solving the problem of localizing the RO, which are the identification, the multi-threshold selection, and the specification of the maximum DF. By modeling the process of forming the DF, numerical estimates have been obtained for the probability of localizing the RO. The research results indicate the feasibility of using the proposed method in a CENS in relation to a VS with several bright objects.

Stability and Performance of the Binary Compensation Unit for Cryogenic Current Comparator Bridges

As a key module of the redesigned cryogenic current comparator setup, the binary compensation unit is now in everyday use for resistance calibration at Physikalisch-Technische Bundesanstalt (PTB), as well as in a number of national metrology institutes. In a first step, its excellent performance was demonstrated in consistency checks involving the quantum Hall resistance and normal resistors with nominal values $100~\Omega $ and 12.9 $\text{k}\Omega $ . Meanwhile, the stability of this compensation unit could be evaluated, too. Our analysis is based on the results of calibrations at regular intervals of 20 weeks covering more than three years. Also the influence of temperature has been studied experimentally. From the data presented here, we conclude that the device under test can be considered to be stable within 1 part in 104 over ten years.

Investigating “concgrams” in the language of contracts and legal agreements

Extended units of meaning have been scarcely investigated thus far in legal phraseology. Moreover, dictionaries of law neglect to include extended versions of prototypical binary units. For instance, nominal consideration is included in the English-Croatian Dictionary of Law (Ga?i? 2010) but its wider context transferred at nil or nominal consideration is not listed. This paper will therefore contrastively investigate extended phraseological units in the language of contracts. The analysis will involve comparing specialized corpora with special purpose (LSP) dictionaries and with reference corpora. The paper proposes that by analysing the extended units of meaning found in contracts and legal agreements one can gain a clearer insight into the natural idiomaticity and formulaicity prevalent in this genre. By comparing the extended units of meaning found in English and Croatian corpora, one can reveal examples of usage of such units with all their associations. The results of such analysis can be applied to translator training, especially where translation is carried out into a second language (L2).

Exergoeconomic analysis and multi-objective optimization of a novel combined flash-binary cycle for Sabalan geothermal power plant in Iran

Employing high-efficiency thermodynamic cycles for power generation from available heat sources is an effective strategy for sustainable energy development. In this regard in the present work, a novel combined flash-binary cycle is proposed for power generation from Sabalan geothermal wells in Iran, considering the wellhead temperature and pressure differences of existing wells. Exergoeconomic approach is applied to investigate the proposed system performance from the viewpoints of both thermodynamics and economics. Using the real data and properties of brine exploited from Sabalan geothermal field, the cycle performance is assessed and a single-objective optimization is performed considering the specific cost of output power as the objective function for four candidate working fluids for binary unit. Then, a multi-objective optimization is conducted and the Pareto frontier as a set of optimal solutions is presented. The results showed that, when single-objective optimization is performed, the proposed cycle with R141b as the binary unit working fluid has the best performance for which the specific cost of output power is calculated to be 4.901$/GJ with an exergy efficiency of 52.56%. However, the multi-objective optimization leads to an exergy efficiency of 54.87% and a power cost of 5.068$/GJ at the selected optimal design point. Finally, a performance comparison is made between the proposed cycle in this work with the previously proposed systems for Sabalan geothermal reservoirs and it is concluded that the flash-binary system proposed here has significantly better performance than the previous systems.

Modern geothermal power: Binary cycle geothermal power plants

In the second part of the review of modern geothermal power plant technologies and equipment, a role, a usage scale, and features of application of binary cycle plants in the geothermal economy are considered. Data on the use of low-boiling fluids, their impact on thermal parameters and performance of geothermal binary power units are presented. A retrospective of the use of various low-boiling fluids in industrial binary power units in the world since 1965 is shown. It is noted that the current generating capacity of binary power units running on hydrocarbons is equal to approximately 82.7% of the total installed capacity of all the binary power units in the world. At the same time over the past 5 years, the total installed capacity of geothermal binary power units in 25 countries increased by more than 50%, reaching nearly 1800 MW (hereinafter electric power is indicated), by 2015. A vast majority of the existing binary power plants recovers heat of geothermal fluid in the range of 100–200°C. Binary cycle power plants have an average unit capacity of 6.3 MW, 30.4 MW at single-flash power plants, 37.4 MW at double-flash plants, and 45.4 MW at power plants working on superheated steam. The largest binary cycle geothermal power plants (GeoPP) with an installed capacity of over 60 MW are in operation in the United States and the Philippines. In most cases, binary plants are involved in the production process together with a steam cycle. Requirements to the fluid ensuring safety, reliability, and efficiency of binary power plants using heat of geothermal fluid are determined, and differences and features of their technological processes are shown. Application of binary cycle plants in the technological process of combined GeoPPs makes it possible to recover geothermal fluid more efficiently. Features and advantages of binary cycle plants using multiple fluids, including a Kalina Cycle, are analyzed. Technical characteristics of binary cycle plants produced by various manufacturers are considered, and data on the Russian pilot binary geothermal power unit in the Pauzhetskaya GeoPP is provided. Expediency of the use of binary cycle plants for autonomous power supply and energy extension of existing GeoPPs without drilling extra wells and in flowsheets of newly designed combined GeoPPs are noted.

Multivariate Time-Series Classification Using the Hidden-Unit Logistic Model.

We present a new model for multivariate time-series classification, called the hidden-unit logistic model (HULM), that uses binary stochastic hidden units to model latent structure in the data. The hidden units are connected in a chain structure that models temporal dependencies in the data. Compared with the prior models for time-series classification such as the hidden conditional random field, our model can model very complex decision boundaries, because the number of latent states grows exponentially with the number of hidden units. We demonstrate the strong performance of our model in experiments on a variety of (computer vision) tasks, including handwritten character recognition, speech recognition, facial expression, and action recognition. We also present a state-of-the-art system for facial action unit detection based on the HULM.

Correlated Fluctuations in Strongly Coupled Binary Networks Beyond Equilibrium

Randomly coupled Ising spins constitute the classical model of collective phenomena in disordered systems, with applications covering ferromagnetism, combinatorial optimization, protein folding, stock market dynamics, and social dynamics. The phase diagram of these systems is obtained in the thermodynamic limit by averaging over the quenched randomness of the couplings. However, many applications require the statistics of activity for a single realization of the possibly asymmetric couplings in finite-sized networks. Examples include reconstruction of couplings from the observed dynamics, learning in the central nervous system by correlation-sensitive synaptic plasticity, and representation of probability distributions for sampling-based inference. The systematic cumulant expansion for kinetic binary (Ising) threshold units with strong, random and asymmetric couplings presented here goes beyond mean-field theory and is applicable outside thermodynamic equilibrium; a system of approximate non-linear equations predicts average activities and pairwise covariances in quantitative agreement with full simulations down to hundreds of units. The linearized theory yields an expansion of the correlation- and response functions in collective eigenmodes, leads to an efficient algorithm solving the inverse problem, and shows that correlations are invariant under scaling of the interaction strengths.

World Geothermal Congress WGC-2015

This article discusses materials and results of the World Geothermal Congress that was held in Melbourne (Australia) from April 19 to April 25, 2015. Information on the extent and technological features of utilization of geothermal resources for heat supply and power production, as well as in other economic areas, is given. A stable growth in the capacity and number of geothermal power systems that is determined by ecological cleanliness, economic efficiency, and the highest (among renewable energy sources) indicators of installed capacity utilization is shown. It was noted that combined schemes of geothermal power plants (GPPs), such as turbine units of different type (binary units, units with one or two separation pressures, etc.), have become more frequently used to increase the efficiency of utilization of geothermal heat carrier. Actual data determining room heating systems with the total worldwide capacity of nearly 50000 MW thermal (MWt) as the most currently significant segment of consumption of geothermal waters are given. In addition, geothermal resources are also utilized in soil pumps, balneological and sports basins, greenhouse complexes, and other manufactures. It was noted that geological studies were carried out in more than 40 countries, with the development of methods of simulation of tanks for the existing and new geothermal fields. Trends of development and the role of geothermal power engineering in the energy supply of many countries are shown. It was shown that prospects for the development of geothermal power generation are significantly associated with utilization of low-temperature geothermal sources in binary power generating units, as well as with the increase in installed capacity of operating geothermal power plants (GPPs) without drilling additional wells, i.e., by using waste geothermal heat carrier in binary-cycle or combined-cycle power plants. The article provides data on a pilot binary power unit at Pauzhetka GPP and on a promising Russian geothermal project to increase the installed capacity of Mutnovsk GPP (whose current capacity is 50.0 (2 × 25.0) MW of electric power) by 25% by constructing a combined binary-cycle power generating unit on the basis of waste separate utilization.

Boltzmann machines with clusters of stochastic binary units

The original restricted Boltzmann machines (RBMs) are extended by replacing the binary visible and hidden variables with clusters of binary units, and a new learning algorithm for training deep Boltzmann machine of this new variant is proposed. The sum of binary units of each cluster is approximated by a Gaussian distribution. Experiments demonstrate that the proposed Boltzmann machines can achieve good performance in the MNIST handwritten digital recognition task.

Combined cycle power unit with a binary system based on waste geothermal brine at Mutnovsk geothermal power plant

The Russian geothermal power systems developed in the last few decades outperform their counterparts around the world in many respects. However, all Russian geothermal power stations employ steam as the geothermal fluid and discard the accompanying geothermal brine. In reality, the power of the existing Russian geothermal power stations may be increased without drilling more wells, if the waste brine is employed in combined cycle systems with steam and binary turbine units. For the example of the 50 MW Mutnovsk geothermal power plant, the optimal combined cycle power unit based on the waste geothermal brine is considered. It is of great interest to determine how the thermodynamic parameters of the secondary steam in the expansion unit and the pressure in the condenser affect the performance of the equipment in the combined cycle power unit at Mutnovsk geothermal power plant. For the utilization of the waste geothermal brine at Mutnovsk geothermal power plant, the optimal air temperature in the condensers of the combined cycle power unit is +5°C. The use of secondary steam obtained by flashing of the geothermal brine at Mutnovsk geothermal power plant 1 at a pressure of 0.2 MPa permits the generation of up to 8 MW of electric power in steam turbines and additional power of 5 MW in the turbines of the binary cycle.

The respective roles of polar/nonpolar binary patterns and amino acid composition in protein regular secondary structures explored exhaustively using hydrophobic cluster analysis.

Several studies have highlighted the leading role of the sequence periodicity of polar and nonpolar amino acids (binary patterns) in the formation of regular secondary structures (RSS). However, these were based on the analysis of only a few simple cases, with no direct mean to correlate binary patterns with the limits of RSS. Here, HCA-derived hydrophobic clusters (HC) which are conditioned binary patterns whose positions fit well those of RSS, were considered. All the HC types, defined by unique binary patterns, which were commonly observed in three-dimensional (3D) structures of globular domains, were analyzed. The 180 HC types with preferences for either α-helices or β-strands distinctly contain basic binary units typical of these RSS. Therefore a general trend supporting the "binary pattern preference" assumption was observed. HC for which observed RSS are in disagreement with their expected behavior (discordant HC) were also examined. They were separated in HC types with moderate preferences for RSS, having "weak" binary patterns and versatile RSS and HC types with high preferences for RSS, having "strong" binary patterns and then displaying nonpolar amino acids at the protein surface. It was shown that in both cases, discordant HC could be distinguished from concordant ones by well-differentiated amino acid compositions. The obtained results could, thus, help to complement the currently available methods for the accurate prediction of secondary structures in proteins from the only information of a single amino acid sequence. This can be especially useful for characterizing orphan sequences and for assisting protein engineering and design.

Globally coupled stochastic two-state oscillators: synchronization of infinite and finite arrays

We consider arrays of the simplest two-state (on–off) stochastic units. The units are Markovian, that is, the transitions between the two states occur at a given rate. We construct arrays of N globally coupled binary units, and observe a remarkable richness of behavior as the control parameter that measures the coupling strength is increased. In the mean field limit as we consider the four simplest polynomial forms of coupling that lead to bifurcations, and characterize the associated phase transitions of the arrays. When N is finite there are fluctuations about the well-defined steady states of the infinite arrays. We study the nature of these fluctuations and their effects on the bifurcations in all cases by constructing the appropriate Langevin equations and the associated Fokker–Planck equations.

All-small-molecule organic solar cells based on an electron donor incorporating binary electron-deficient units

A conjugated small molecule incorporating binary electron-deficient units was applied in all-small-molecule organic solar cells using fullerene or non-fullerene derivatives as the electron acceptor.

Electronic health record analysis via deep Poisson factor models

Electronic Health Record (EHR) phenotyping utilizes patient data captured through normal medical practice, to identify features that may represent computational medical phenotypes. These features may be used to identify at-risk patients and improve prediction of patient morbidity and mortality. We present a novel deep multi-modality architecture for EHR analysis (applicable to joint analysis of multiple forms of EHR data), based on Poisson Factor Analysis (PFA) modules. Each modality, composed of observed counts, is represented as a Poisson distribution, parameterized in terms of hidden binary units. Information from different modalities is shared via a deep hierarchy of common hidden units. Activation of these binary units occurs with probability characterized as Bernoulli-Poisson link functions, instead of more traditional logistic link functions. In addition, we demonstrate that PFA modules can be adapted to discriminative modalities. To compute model parameters, we derive efficient Markov Chain Monte Carlo (MCMC) inference that scales efficiently, with significant computational gains when compared to related models based on logistic link functions. To explore the utility of these models, we apply them to a subset of patients from the Duke-Durham patient cohort. We identified a cohort of over 16,000 patients with Type 2 Diabetes Mellitus (T2DM) based on diagnosis codes and laboratory tests out of our patient population of over 240,000. Examining the common hidden units uniting the PFA modules, we identify patient features that represent medical concepts. Experiments indicate that our learned features are better able to predict mortality and morbidity than clinical features identified previously in a large-scale clinical trial.

Area Efficient and Fast Combined Binary/Decimal Floating Point Fused Multiply Add Unit

In this work we present a new 64-bit floating point Fused Multiply Add (FMA) unit that can perform both binary and decimal addition, multiplication, and fused-multiply-add operations. The presented FMA has 6 percent less delay than the fastest stand-alone decimal unit and 23 percent less area than both binary and decimal units together. These results were achieved by the use of: 1) column by column reduction to reduce the partial products in the multiplier tree, 2) a new leading zeros detector that produces its output in base-3 to simplify the normalization shifting in the binary datapath, 3) the use of a redundant adder to perform the final addition, 4) using a new rounding-while-redundant technique to hide the rounding delay and remove it from the critical path, and 5) using a new simple conversion technique from redundant to binary/decimal.

MATHEMATICAL FOUNDATION OF BINARY SYSTEM ONTOLOGY FOR DATA PROCESSING

The paper deals with theoretical principles that guarantee Boolean functional completeness of some primitive types of heterogeneous and homogeneous binary operational units designed for a logical data processing. Information in such units can be represented by various (different) physical parameters coding different states of physical objects. Simplest functionalities that define output “signals” (“output” parameters) as primitive functions of input “signals” (“input” parameters) are analyzed. It is shown that heterogeneous and homogeneous binary systems for data processing can be constructed on the basis of a general concept using proper object models based on General Formal Technological approach. It is shown that the homogeneous and heterogeneous binary information processing systems can be constructed based on the total (universal) basic ontology, which considers all physical objects as functional convert a certain set of physical parameters (physical properties in its broadest sense) in the other.

A Case Study on Medium-Term Hydropower Scheduling with Sales of Capacity

This paper conducts a case study on hydropower scheduling considering sales of capacity reserves and impacts of detailed mod-elling. In latter years a growing demand for reserve capacity has been needed to ensure stable operation in the power grid which has given the power producers incentives to commence methods for co-optimization of energy and capacity. In this paper we will assess the value of providing primary reserves and how decisive accurate modelling is for sale of capacity. The results are based on a model consisting of a Strategy and a Simulator part. The Strategy Model is based on a combined Stochastic Dynamic Pro-gramming (SDP)/Stochastic Dual Dynamic Programming (SDDP) model where variables and functional relationships are linear. Subsequently the obtained profit-to-go function is used in the Simulator Model; a Mixed Integer Program (MIP) based simulator allowing a more detailed system description. The case study represents a Norwegian water course comprising of two minor and a large regulating reservoir. The Simulator Model manages to incorporate a binary unit commitment and to represent the non-convex relationship between power and discharge, giving more viable results and identifying dependencies between the reservoirs. As a result it was found that the expected profit from sales of capacity was reduced by 40% when comparing results from the Simulator with the Strategy Model. Correspondingly, sales of capacity gave a clear shift of power outputs between the models as unrealistic results were eliminated in the Simulator Model.