Machine Olfaction Research Articles

Discrimination of Chinese Herbal Medicine by Machine Olfaction

“Small Sample Size” (SSS) problem would occur while using linear discriminant analysis (LDA) algorithm with traditional Fisher criterion if the within-class scatter matrix is singular. The combination of maximum scatter difference (MSD) criterion and LDA algorithm for solve SSS problem is described. It is employed to detect three kinds of Chinese herbal medicines from different growing areas by machine olfaction. Compared with PCA or PCA + LDA algorithm, the classification result was enhanced. It works out that only a few samples of Anhui Atractylodes are classified incorrectly, however, the classification rate reaches 97.8%. DOI: http://dx.doi.org/10.11591/telkomnika.v11i2.1986

Recognizing Sights, Smells, and Sounds with Gnostic Fields

Mammals rely on vision, audition, and olfaction to remotely sense stimuli in their environment. Determining how the mammalian brain uses this sensory information to recognize objects has been one of the major goals of psychology and neuroscience. Likewise, researchers in computer vision, machine audition, and machine olfaction have endeavored to discover good algorithms for stimulus classification. Almost 50 years ago, the neuroscientist Jerzy Konorski proposed a theoretical model in his final monograph in which competing sets of “gnostic” neurons sitting atop sensory processing hierarchies enabled stimuli to be robustly categorized, despite variations in their presentation. Much of what Konorski hypothesized has been remarkably accurate, and neurons with gnostic-like properties have been discovered in visual, aural, and olfactory brain regions. Surprisingly, there have not been any attempts to directly transform his theoretical model into a computational one. Here, I describe the first computational implementation of Konorski's theory. The model is not domain specific, and it surpasses the best machine learning algorithms on challenging image, music, and olfactory classification tasks, while also being simpler. My results suggest that criticisms of exemplar-based models of object recognition as being computationally intractable due to limited neural resources are unfounded.

Classification of Chinese Herbal Medicine Based on Improved LDA Algorithm Using Machine Olfaction

Linear discriminant analysis (LDA) is a popular method among pattern recognition algorithms of machine olfaction. However, “Small Sample Size” (SSS) problem would occur while using LDA algorithm with traditional Fisher criterion if the within-class scatter matrix is singular. In this paper, maximum scatter difference (MSD) criterion and LDA were combined to solve SSS problem, so that three kinds of Chinese herbal medicines from different growing areas were accurately classified. At the same time, the classification result was enhanced. It works out that only a few samples of Anhui Atractylodes are classified incorrectly, however, the classification rate reaches 97.8%.

Guest Editorial - Special issue on machine olfaction

This year is the 10th anniversary of the first special issue on machine olfaction of the IEEE SENSORS JOURNAL. During this past decade, the field has grown in a number of exciting directions, including spectroscopic and olfactory receptorbased sensing, computational models of olfactory processing, and mobile and distributed sensing. This second special issue provides a timely update on advances during the past decade (as well as a vantage point from which to evaluate the last 30 years) in the field and, more importantly, the challenges that still lie ahead. It includes invited papers based upon presentations give at the International Symposium on Olfaction and Chemical Sensing (ISOCS) that was held in New York Digital Object Identifier 10.1109/JSEN.2012.2215434 City in 2011 (see www.olfactionsociety.org). To provide a balanced view of the various advances in the field, the special issue is organized into six distinct topic areas: biological sensors, chemical sensors, sensor systems, data processing, and electronic nose applications. The topic area of biological sensors recognises the fact that considerable effort is being directed towards improving the capability of artificial olfactory sensors by exploring more closely the biological system and seeking to mimic it more closely.

Signal and Data Processing for Machine Olfaction and Chemical Sensing: A Review

Signal and data processing are essential elements in electronic noses as well as in most chemical sensing instruments. The multivariate responses obtained by chemical sensor arrays require signal and data processing to carry out the fundamental tasks of odor identification (classification), concentration estimation (regression), and grouping of similar odors (clustering). In the last decade, important advances have shown that proper processing can improve the robustness of the instruments against diverse perturbations, namely, environmental variables, background changes, drift, etc. This article reviews the advances made in recent years in signal and data processing for machine olfaction and chemical sensing.

A software tool for large-scale synthetic experiments based on polymeric sensor arrays

Abstract This manuscript introduces a software tool that allows for the design of synthetic experiments in machine olfaction. The proposed software package includes both, a virtual sensor array that reproduces the diversity and response of a polymer array and tools for data generation. The synthetic array of sensors allows for the generation of chemosensor data with a variety of characteristics: unlimited number of sensors, support of multicomponent gas mixtures and full parametric control of the noise in the system. The artificial sensor array is inspired from a reference database of seventeen polymeric sensors with concentration profiles for three analytes. The main features in the sensor data, like sensitivity, diversity, drift and sensor noise, are captured by a set of models under simplified assumptions. The generator of sensor signals can be used in applications related to test and benchmarking of signal processing methods, neuromorphic simulations in machine olfaction and educational tools. The software is implemented in R language and can be freely accessed at: http://chemosensors.r-forge.r-project.org/ .

On time series features and kernels for machine olfaction

Abstract Kernel methods such as support vector machines are a powerful technique to solve pattern recognition problems. One of the important properties of kernel methods is that they can be applied to any kind of input domain, for which it is possible to construct an appropriate kernel. Over the past years, there has been a tremendous interest and progress in the machine learning community to design kernels for “non-standard” data sets, i.e., for data without a vectorial feature representation; examples include graphs, strings, trees, and other such discrete objects. In this paper, we investigate the benefit of using time series kernels to solve machine olfaction applications. In particular, we apply these time series kernels for two pattern recognition problems in machine olfaction, namely, odor classification and odor localization in an open sampling system. We also study the use of time series feature extraction methods, in which features are extracted by making assumptions on the underlying mechanism that generate the time series. Experimental results clearly indicate the advantage of using these kernels when compared to naive techniques that discard the temporal information in the data, and, even more interestingly, these kernels also perform better that techniques that rely on an explicit feature extraction step prior to solving the pattern recognition problem.

Bio-inspired solutions to the challenges of chemical sensing

Bio-inspired solutions to the challenges of chemical sensing

Characterization of a clinical olfactory test with an artificial nose

Clinical olfactory tests are used to address hyposmia/anosmia levels in patients with different types of olfactory impairments. Typically, a given test is employed clinically and then replaced by a new one after a certain period of use which can range from days to several months. There is a need to assess control quality of these tests and also for a procedure to quantify their degradation over time. In this paper we propose a protocol to employ low-cost artificial noses for the quantitative characterization of olfactory tests used in clinical studies. In particular, we discuss a preliminary study on the Connecticut Chemosensorial Clinical Research Center Test kit which shows that some odorants, as sensed by an artificial nose, seem to degrade while others are potentiated as the test ages. We also discuss the need to establish a map of correspondence between human and machine olfaction when artificial noses are used to characterize or compare human smell performance in research and clinical studies.

Sensor selection and chemo-sensory optimization: toward an adaptable chemo-sensory system.

Over the past two decades, despite the tremendous research on chemical sensors and machine olfaction to develop micro-sensory systems that will accomplish the growing existent needs in personal health (implantable sensors), environment monitoring (widely distributed sensor networks), and security/threat detection (chemo/bio warfare agents), simple, low-cost molecular sensing platforms capable of long-term autonomous operation remain beyond the current state-of-the-art of chemical sensing. A fundamental issue within this context is that most of the chemical sensors depend on interactions between the targeted species and the surfaces functionalized with receptors that bind the target species selectively, and that these binding events are coupled with transduction processes that begin to change when they are exposed to the messy world of real samples. With the advent of fundamental breakthroughs at the intersection of materials science, micro- and nano-technology, and signal processing, hybrid chemo-sensory systems have incorporated tunable, optimizable operating parameters, through which changes in the response characteristics can be modeled and compensated as the environmental conditions or application needs change. The objective of this article, in this context, is to bring together the key advances at the device, data processing, and system levels that enable chemo-sensory systems to “adapt” in response to their environments. Accordingly, in this review we will feature the research effort made by selected experts on chemical sensing and information theory, whose work has been devoted to develop strategies that provide tunability and adaptability to single sensor devices or sensory array systems. Particularly, we consider sensor-array selection, modulation of internal sensing parameters, and active sensing. The article ends with some conclusions drawn from the results presented and a visionary look toward the future in terms of how the field may evolve.

Special issue on machine olfaction

Special issue on machine olfaction

Special issue on machine olfaction

Special issue on machine olfaction

Odour detection methods: olfactometry and chemical sensors.

The complexity of the odours issue arises from the sensory nature of smell. From the evolutionary point of view olfaction is one of the oldest senses, allowing for seeking food, recognizing danger or communication: human olfaction is a protective sense as it allows the detection of potential illnesses or infections by taking into account the odour pleasantness/unpleasantness. Odours are mixtures of light and small molecules that, coming in contact with various human sensory systems, also at very low concentrations in the inhaled air, are able to stimulate an anatomical response: the experienced perception is the odour. Odour assessment is a key point in some industrial production processes (i.e., food, beverages, etc.) and it is acquiring steady importance in unusual technological fields (i.e., indoor air quality); this issue mainly concerns the environmental impact of various industrial activities (i.e., tanneries, refineries, slaughterhouses, distilleries, civil and industrial wastewater treatment plants, landfills and composting plants) as sources of olfactory nuisances, the top air pollution complaint. Although the human olfactory system is still regarded as the most important and effective “analytical instrument” for odour evaluation, the demand for more objective analytical methods, along with the discovery of materials with chemo-electronic properties, has boosted the development of sensor-based machine olfaction potentially imitating the biological system. This review examines the state of the art of both human and instrumental sensing currently used for the detection of odours. The olfactometric techniques employing a panel of trained experts are discussed and the strong and weak points of odour assessment through human detection are highlighted. The main features and the working principles of modern electronic noses (E-Noses) are then described, focusing on their better performances for environmental analysis. Odour emission monitoring carried out through both the techniques is finally reviewed in order to show the complementary responses of human and instrumental sensing.

Performance of Machine Olfaction: Effect of Uniqueness of the Initial Data and Information Coding on the Discrimination Ability of Multisensor Arrays

The optimization of a sensor array for a concrete analytical task is usually concerned with choosing a set of sensors to provide the best classification. In this work, a method for the prediction of the quality of classification by evaluation of the uniqueness of the raw experimental data is proposed. The key feature of the method is the presentation of the response of array as a function of the responses of its sensors. The dispersion of those functions serves as quantitative measure of uniqueness of the experimental data for a given set of analytes. The efficiency of the approach has been successfully demonstrated using both simulated and experimental data obtained from the array of three mass-sensitive sensors. The best conformity of the classification efficiency in cluster analysis with results obtained in the framework of the proposed approach is observed in the case of Langmuir-type adsorption processes.

Predicting odor mixture's responses on machine olfaction sensors

One of the challenging issues in current research on machine olfaction devices, which are often called electronic noses (e-noses), is how to approximate or predict the sensor response to odor mixtures. When each odor is produced by its own unique set of odorant compounds, combinations of these unique odorant sets create a sensing challenge for the e-noses with a limited number of elements in its sensing array. One possible approach proposed in the literature is based on an “additive law of mixing” model but it fails in a complex odor mixtures. Another method adopted a specific hardware solution called odor recorder developed by using active odor sensing system. In this study, signal decomposition/reconstruction based on wavelet analysis and support vector regression are adopted to predict a sensor's response to mixtures of odors. The prediction results of our method are investigated and compared with the real sensor responses collected from a commercial e-nose machine, the AppliedSensor NST 3320. We find that the proposed method provides good prediction when applied to different mixing ratios of some coffees and green tea.

Intelligent method for sensor subset selection for machine olfaction

A fundamental design concept for an array of sensors used in machine olfaction devices, electronic noses (e-noses), is that each sensor should maximize the overall sensitivity and provides different selectivity profiles over the range of target odor application. Ideally, each sensor should produce a different response to a given odor so that a unique odor pattern is created. Since this is rarely the case, sensor selection or reduction is needed when classification performance, cost, and technology limitations are issues of concern. The first step in the selection/reduction process is to generate features from each sensor's output waveform. In practice, some of the features obtained from an array of sensors are redundant and irrelevant due to cross-sensitivity and odor characteristics. As a result, inappropriate features or a poor configuration of features can lead to a deterioration of classification performance, or a more complex classification algorithm may be required. Hence, sensor selection for e-nose systems is of great important. In this study, a novel computationally efficient method is introduced by selecting the first few critical sensors based on a maximum margin criterion among different odor classes. Then, a stochastic search algorithm, a genetic algorithm (GA), uses those features as an initial step to optimize our sensor selection problem. The advantages of the proposed method are not only to avoid any initial misstep starting the search, but also to reduce the searching space for the optimal sensor array. From the experimental results on coffee and soda data sets, the number of selected sensors is significantly reduced (up to 90%) and classification performance is near 100%.

Neuro-based olfactory model for artificial organoleptic tests

Recently, the demand for odor processing apparatus in the fragrance and food industries has increased. In this article, we construct a neural network model of the olfactory system as the basis for artificial organoleptic tests that combine the advantages of both human sensory evaluation and machine olfaction. The simulation results indicate that the model can predict odor coding on the glomeruli by appropriately adjusting the parameters involved. Further, the model can simulate the feature extraction ability known as “attention”.

Meat Quality Assessment by Electronic Nose (Machine Olfaction Technology)

Over the last twenty years, newly developed chemical sensor systems (so called “electronic noses”) have made odor analyses possible. These systems involve various types of electronic chemical gas sensors with partial specificity, as well as suitable statistical methods enabling the recognition of complex odors. As commercial instruments have become available, a substantial increase in research into the application of electronic noses in the evaluation of volatile compounds in food, cosmetic and other items of everyday life is observed. At present, the commercial gas sensor technologies comprise metal oxide semiconductors, metal oxide semiconductor field effect transistors, organic conducting polymers, and piezoelectric crystal sensors. Further sensors based on fibreoptic, electrochemical and bi-metal principles are still in the developmental stage. Statistical analysis techniques range from simple graphical evaluation to multivariate analysis such as artificial neural network and radial basis function. The introduction of electronic noses into the area of food is envisaged for quality control, process monitoring, freshness evaluation, shelf-life investigation and authenticity assessment. Considerable work has already been carried out on meat, grains, coffee, mushrooms, cheese, sugar, fish, beer and other beverages, as well as on the odor quality evaluation of food packaging material. This paper describes the applications of these systems for meat quality assessment, where fast detection methods are essential for appropriate product management. The results suggest the possibility of using this new technology in meat handling.

Sensor Selection for Machine Olfaction Based on Transient Feature Extraction

Machine olfaction devices, which are often called electronic noses (e-noses), are gaining favor for odor assessment applications in several industrial sectors, such as beverage, perfumery, and food. From a design point of view, the number of sensors in these devices for a particular odor application should be minimized without degrading classification accuracy. This paper deals with selecting sensors for e-noses to make small portable devices with fast response times and reduced cost possible. Prior research efforts have been reported in the open literature and have shown that many advantages can be gained by properly selecting the input features before forwarding to a pattern classification algorithm. This selection process can reduce the dimensionality of the feature space, remove redundant and irrelevant features, speed up classification, and improve classification performance. In this paper, the transient features of an array of sensors obtained by applying a multiresolutional approximation technique from the discrete wavelet transform (DWT) are investigated to search for an optimal sensor array to be implemented in the e-nose system. A genetic algorithm is adapted to tailor a gas sensor array for two different odor data sets (coffee and soda). From the experimental results, the input features obtained by applying the DWT to the transient sensor responses not only provide a significant reduction in the number of sensors when compared to traditional features but also improve the classification rate to near 100%.

Towards an artificial olfactory mucosa for improved odour classification

Here, we report on a biologically inspired analytical system that represents a new concept in the field of machine olfaction. Specifically, this paper describes the design and fabrication of a novel sensor system, based upon the principle of ‘nasal chromatography’, which emulates the human olfactory mucosa. Our approach exploits the physical positioning of a series of broadly tuned sensors (equivalent to the olfactory epithelium) along the length of a planar chromatographic channel (analogous to the thin mucus coating of the nasal cavity) from which we extract both spatial (response magnitude) and temporal (retentive delay) sensor signals. Our study demonstrates that this artificial mucosa is capable of generating both spatial and temporal signals which, when combined, create a novel spatio-temporal representation of an odour. We believe that such a system not only offers improved odour discrimination over a sensor array-based electronic nose, but also shorter analysis times than conventional gas chromatographic techniques.