Forrest Sheng Bao

A Leaf Recognition Algorithm for Plant Classification Using Probabilistic Neural Network

In this paper, we employ probabilistic neural network (PNN) with image and data processing techniques to implement a general purpose automated leaf recognition for plant classification. 12 leaf features are extracted and orthogonalized into 5 principal variables which consist the input vector of the PNN. The PNN is trained by 1800 leaves to classify 32 kinds of plants with an accuracy greater than 90%. Compared with other approaches, our algorithm is an accurate artificial intelligence approach which is fast in execution and easy in implementation.

Small RNA Deep Sequencing Reveals Role for Arabidopsis thaliana RNA-Dependent RNA Polymerases in Viral siRNA Biogenesis

RNA silencing functions as an important antiviral defense mechanism in a broad range of eukaryotes. In plants, biogenesis of several classes of endogenous small interfering RNAs (siRNAs) requires RNA-dependent RNA Polymerase (RDR) activities. Members of the RDR family proteins, including RDR1and RDR6, have also been implicated in antiviral defense, although a direct role for RDRs in viral siRNA biogenesis has yet to be demonstrated. Using a crucifer-infecting strain of Tobacco Mosaic Virus (TMV-Cg) and Arabidopsis thaliana as a model system, we analyzed the viral small RNA profile in wild-type plants as well as rdr mutants by applying small RNA deep sequencing technology. Over 100,000 TMV-Cg-specific small RNA reads, mostly of 21- (78.4%) and 22-nucleotide (12.9%) in size and originating predominately (79.9%) from the genomic sense RNA strand, were captured at an early infection stage, yielding the first high-resolution small RNA map for a plant virus. The TMV-Cg genome harbored multiple, highly reproducible small RNA-generating hot spots that corresponded to regions with no apparent local hairpin-forming capacity. Significantly, both the rdr1 and rdr6 mutants exhibited globally reduced levels of viral small RNA production as well as reduced strand bias in viral small RNA population, revealing an important role for these host RDRs in viral siRNA biogenesis. In addition, an informatics analysis showed that a large set of host genes could be potentially targeted by TMV-Cg-derived siRNAs for posttranscriptional silencing. Two of such predicted host targets, which encode a cleavage and polyadenylation specificity factor (CPSF30) and an unknown protein similar to translocon-associated protein alpha (TRAP α), respectively, yielded a positive result in cleavage validation by 5′RACE assays. Our data raised the interesting possibility for viral siRNA-mediated virus-host interactions that may contribute to viral pathogenicity and host specificity.

Hierarchical macrotube/mesopore carbon decorated with mono-dispersed Ag nanoparticles as a highly active catalyst

Natural wood, featuring abundant oxygen-containing functional groups, has been utilized as a reductant to synthesize monodispersed Ag nanoparticles on its surface. By further carbonization of the Ag/wood composite, wood was converted to carbon with embedded mesopore structures. Through the two-step reduction and carbonization, a macro-tube/meso-pore carbon frame with decorated mono-dispersed silver nanoparticles (Ag/C) can be conveniently synthesized. Various characterization techniques including SEM, TEM, HRTEM, BET, Raman, XRD, XPS and FT-IR have been utilized to study the material microstructure, crystalline structure, pore size and surface area and surface properties. The mechanism of Ag/wood formation has also been studied in this work. Ag/C shows outstanding activity in 4-nitrophenol and 2-nitrophenol reduction reactions with much higher reaction rate than literature reports, and no obvious activity degradation was observed after 10 cycles of durability tests. This newly developed synthetic methodology could serve as a general tool to design and synthesize other metal/carbon nanocomposite catalysts for a wider range of catalytic applications. More importantly, the utilization of a widely accessible renewable resource provides a sustainable feature of this work to reduce manufacturing cost and environmental impact.

Automated epilepsy diagnosis using interictal scalp EEG

Over 50 million people worldwide suffer from epilepsy. Traditional diagnosis of epilepsy relies on tedious visual screening by highly trained clinicians from lengthy EEG recording that contains the presence of seizure (ictal) activities. Nowadays, there are many automatic systems that can recognize seizure-related EEG signals to help the diagnosis. However, it is very costly and inconvenient to obtain long-term EEG data with seizure activities, especially in areas short of medical resources. We demonstrate in this paper that we can use the interictal scalp EEG data, which is much easier to collect than the ictal data, to automatically diagnose whether a person is epileptic. In our automated EEG recognition system, we extract three classes of features from the EEG data and build Probabilistic Neural Networks (PNNs) fed with these features. We optimize the feature extraction parameters and combine these PNNs through a voting mechanism. As a result, our system achieves an impressive 94.07% accuracy.

Semantic Analysis and Helpfulness Prediction of Text for Online Product Reviews

Predicting the helpfulness of product reviews is a key component of many ecommerce tasks such as review ranking and recommendation. However, previous work mixed review helpfulness prediction with those outer layer tasks. Using nontext features, it leads to less transferable models. This paper solves the problem from a new angle by hypothesizing that helpfulness is an internal property of text. Purely using review text, we isolate review helpfulness prediction from its outer layer tasks, employ two interpretable semantic features, and use human scoring of helpfulness as ground truth. Experimental results show that the two semantic features can accurately predict helpfulness scores and greatly improve the performance compared with using features previously used. Cross-category test further shows the models trained with semantic features are easier to be generalized to reviews of different product categories. The models we built are also highly interpretable and align well with human annotations.

Rapid Prediction of Bacterial Heterotrophic Fluxomics Using Machine Learning and Constraint Programming

13C metabolic flux analysis (13C-MFA) has been widely used to measure in vivo enzyme reaction rates (i.e., metabolic flux) in microorganisms. Mining the relationship between environmental and genetic factors and metabolic fluxes hidden in existing fluxomic data will lead to predictive models that can significantly accelerate flux quantification. In this paper, we present a web-based platform MFlux (http://mflux.org) that predicts the bacterial central metabolism via machine learning, leveraging data from approximately 100 13C-MFA papers on heterotrophic bacterial metabolisms. Three machine learning methods, namely Support Vector Machine (SVM), k-Nearest Neighbors (k-NN), and Decision Tree, were employed to study the sophisticated relationship between influential factors and metabolic fluxes. We performed a grid search of the best parameter set for each algorithm and verified their performance through 10-fold cross validations. SVM yields the highest accuracy among all three algorithms. Further, we employed quadratic programming to adjust flux profiles to satisfy stoichiometric constraints. Multiple case studies have shown that MFlux can reasonably predict fluxomes as a function of bacterial species, substrate types, growth rate, oxygen conditions, and cultivation methods. Due to the interest of studying model organism under particular carbon sources, bias of fluxome in the dataset may limit the applicability of machine learning models. This problem can be resolved after more papers on 13C-MFA are published for non-model species.

ASP+POMDP: Integrating non-monotonic logic programming and probabilistic planning on robots

Mobile robots equipped with multiple sensors and deployed in real-world domains frequently find it difficult to process all sensor inputs, or to operate without any human input and domain knowledge. At the same time, robots cannot be equipped with all relevant domain knowledge in advance, and humans are unlikely to have the time and expertise to provide elaborate and accurate feedback. This paper presents a novel framework that addresses these challenges by integrating high-level logical inference with low-level probabilistic sequential decision-making. Specifically, Answer Set Programming (ASP), a non-monotonic logic programming paradigm, is used to represent, reason with and revise domain knowledge obtained from sensor inputs and high-level human feedback, while hierarchical partially observable Markov decision processes (POMDPs) are used to automatically adapt visual sensing and information processing to the task at hand. Furthermore, a psychophysics-inspired strategy is used to merge the output of logical inference with probabilistic beliefs. All algorithms are evaluated in simulation and on wheeled robots localizing target objects in indoor domains.

A high-performance seizure detection algorithm based on Discrete Wavelet Transform (DWT) and EEG

In the past decade, Discrete Wavelet Transform (DWT), a powerful time-frequency tool, has been widely used in computer-aided signal analysis of epileptic electroencephalography (EEG), such as the detection of seizures. One of the important hurdles in the applications of DWT is the settings of DWT, which are chosen empirically or arbitrarily in previous works. The objective of this study aimed to develop a framework for automatically searching the optimal DWT settings to improve accuracy and to reduce computational cost of seizure detection. To address this, we developed a method to decompose EEG data into 7 commonly used wavelet families, to the maximum theoretical level of each mother wavelet. Wavelets and decomposition levels providing the highest accuracy in each wavelet family were then searched in an exhaustive selection of frequency bands, which showed optimal accuracy and low computational cost. The selection of frequency bands and features removed approximately 40% of redundancies. The developed algorithm achieved promising performance on two well-tested EEG datasets (accuracy >90% for both datasets). The experimental results of the developed method have demonstrated that the settings of DWT affect its performance on seizure detection substantially. Compared with existing seizure detection methods based on wavelet, the new approach is more accurate and transferable among datasets.

A REVIEW OF TREE CONVEX SETS TEST

A collection of sets may have some interesting properties which help identify efficient algorithms for constraint satisfaction problems and combinatorial auction problems. One of the properties is tree convexity. A collection S of sets is tree convex if we can find a tree T whose nodes are the union of the sets of S and each set of S is the nodes of a subtree of T. This concept extends that of row convex sets each of which is an interval over a total ordering of the elements of the union of these sets. An interesting problem is to find efficient algorithms to test whether a collection of sets is tree convex. It is not known before if there exists a linear time algorithm for this test. In this paper, we review the materials that are the key to a linear algorithm: hypergraphs, a characterization of tree convex sets and the acyclic hypergraph test algorithm. Some typos in the original paper of the acyclicity test are corrected here. Some experiments show that the linear algorithm is significantly faster than a well‐known existing algorithm.

Epileptic focus localization using EEG based on discrete wavelet transform through full-level decomposition

Electroencephalogram (EEG) is a gold standard in epilepsy diagnosis and has been widely studied for epilepsy-related signal classification, such as seizure detection or focus localization. In the past few years, discrete wavelet transform (DWT) has been widely used to analyze epileptic EEG. However, one practical question unanswered is the optimal levels of wavelet decomposition. Deeper DWT can yield a more detailed depiction of signals but it requires substantially more computational time. In this paper, we study this problem, using the most difficult epileptic EEG classification task, focus localization, as an example. The results show that decomposition level effects the localization accuracy more significantly than mother wavelets. For all wavelets, decomposition beyond level 7 improves accuracy limitedly and even decreases accuracy. We further study what are the most effective bands and features for focus localization. An interpretation of our results is that focal and non-focal epileptic EEGs differ the most at high frequencies of EEG rhythms. The best accuracy of epileptic focus localization achieved in this research is 83.07% using sym6 from levels 1 to 7.