Data-Driven Channel Pruning towards Local Binary Convolution Inverse Bottleneck Network Based on Squeeze-and-Excitation Optimization Weights

This paper proposed a model pruning method based on local binary convolution (LBC) and squeeze-and-excitation (SE) optimization weights. We first proposed an efficient deep separation convolution model based on the LBC kernel. By expanding the number of LBC kernels in the model, we have trained a larger model with better results, but more parameters and slower calculation speed. Then, we extract the SE optimization weight value of each SE module according to the data samples and score the LBC kernel accordingly. Based on the score of each LBC kernel corresponding to the convolution channel, we performed channel-based model pruning, which greatly reduced the number of model parameters and accelerated the calculation speed. The model pruning method proposed in this paper is verified in the image classification database. Experiments show that, in the model using the LBC kernel, as the number of LBC kernels increases, the recognition accuracy will increase. At the same time, the experiment also proved that the recognition accuracy is maintained at a similar level in the small parameter model after channel-based model pruning by the SE optimization weight value.

A CONVblock for Convolutional Neural Networks

The reduction in the size of convolution filters has been shown to be effective in image classification models. They make it possible to reduce the calculation and the number of parameters used in the operations of the convolution layer while increasing the efficiency of the representation. The authors present a deep architecture for classification with improved performance. The main objective of this architecture is to improve the main performances of the network thanks to a new design based on CONVblock. The proposal is evaluated on a classification database: CIFAR-10 and MNIST. The experimental results demonstrate the effectiveness of the proposed method. This architecture offers an error of 1.4% on CIFAR-10 and 0.055% on MNIST.

The Transporter Classification Database (TCDB): 2021 update

The Transporter Classification Database (TCDB; tcdb.org) is a freely accessible reference resource, which provides functional, structural, mechanistic, medical and biotechnological information about transporters from organisms of all types. TCDB is the only transport protein classification database adopted by the International Union of Biochemistry and Molecular Biology (IUBMB) and now (October 1, 2020) consists of 20 653 proteins classified in 15 528 non-redundant transport systems with 1567 tabulated 3D structures, 18 336 reference citations describing 1536 transporter families, of which 26% are members of 82 recognized superfamilies. Overall, this is an increase of over 50% since the last published update of the database in 2016. This comprehensive update of the database contents and features include (i) adoption of a chemical ontology for substrates of transporters, (ii) inclusion of new superfamilies, (iii) a domain-based characterization of transporter families for the identification of new members as well as functional and evolutionary relationships between families, (iv) development of novel software to facilitate curation and use of the database, (v) addition of new subclasses of transport systems including 11 novel types of channels and 3 types of group translocators and (vi) the inclusion of many man-made (artificial) transmembrane pores/channels and carriers.

Deep-sea Coral Taxonomy Workshop, Colombia 2019

Corals are some of the conspicuous taxa in deep-sea ecosystems. Yet, characterizing coral diversity is difficult and requires a combination of both morphological and genetic data. Many leading coral taxonomy experts are close-to retirement or have already retired. It is now imperative that the hands-on expertise that these taxonomists have – much of which is not captured in manuscripts or books – is transferred to the next generation. The Deep-Sea Coral Taxonomy Workshop, funded by a Lounsbery award from the Deep-Sea Biology Society, aimed to provide a training opportunities and build taxonomic capacity in Colombia and Latin America. Workshop participants examined the deep-sea coral diversity of the southern western Caribbean, a poorly explored region. The three-day workshop was based mainly on hands-on activities focused on octocorals and black corals, and included introductory talks to the taxonomy of these groups and identification activities using specimens. Thanks to the workshop, it was possible to review and update the classification database of the Makuriwa Marine Natural History Museum collection. Additionally, four new species from the families Clavulariidae, Plexauridae and Gorgoniidae were identified and will be described in the near future.

Support Vector Feature Extraction Based Lasso For Gender Recognition From Object Classification

Image processing is a field in which biometric traits such as Face, voice, lip movements, hand geometry, odour, gait, iris, retina, fingerprint etc., are essential for recognition. The face is the most critical biometric trait for recognition because the face is an easily approachable biometric trait. There is no need for attention from a human being for face recognition. Human face classification is a challenging task for a machine. In this project, minimum distance classifier used with LASSO based gender classification. Database of 100 images (50 male and 50 female face images which considered from 4 different databases) used for face recognition and classification. Original face image database used for the gender classification. This approach of dual classfication ((1) Recognizing or classfying human faces from various objects and (2) Classifying gender through face recognition) is made possible with the help of combining modified SIFT feature in combination with ridge regression (RR), elastic net (EN) and lasso regression with GSVM (LRGS) based classificatioins. The final classfication results with accuracy are as follows for RR- 89.6%, EN- 93.5%, LR-93.2% and propsed approach(LRGS) with 98.4% accurate detection rate with rediction names.

CLRS: Continual Learning Benchmark for Remote Sensing Image Scene Classification

Remote sensing image scene classification has a high application value in the agricultural, military, as well as other fields. A large amount of remote sensing data is obtained every day. After learning the new batch data, scene classification algorithms based on deep learning face the problem of catastrophic forgetting, that is, they cannot maintain the performance of the old batch data. Therefore, it has become more and more important to ensure that the scene classification model has the ability of continual learning, that is, to learn new batch data without forgetting the performance of the old batch data. However, the existing remote sensing image scene classification datasets all use static benchmarks and lack the standard to divide the datasets into a number of sequential learning training batches, which largely limits the development of continual learning in remote sensing image scene classification. First, this study gives the criteria for training batches that have been partitioned into three continual learning scenarios, and proposes a large-scale remote sensing image scene classification database called the Continual Learning Benchmark for Remote Sensing (CLRS). The goal of CLRS is to help develop state-of-the-art continual learning algorithms in the field of remote sensing image scene classification. In addition, in this paper, a new method of constructing a large-scale remote sensing image classification database based on the target detection pretrained model is proposed, which can effectively reduce manual annotations. Finally, several mainstream continual learning methods are tested and analyzed under three continual learning scenarios, and the results can be used as a baseline for future work.

Protein structure featurization via standard image classification neural networks

ABSTRACTMany applications in the biomedical domain involve the detailed molecular and functional characterization of macro-molecules such as proteins. Where possible, this involves the knowledge of detailed 3D coordinates of every atom within a protein. At the same time, machine learning has become the basis of much innovation within this domain in recent years. There are, however, a few challenges in applying machine learning to 3D protein structures, such as variability in size and high dimensionality of the data. It would therefore be beneficial to be able to map every protein structure to a smaller fixed-dimensional representation that is directly learned from the structure without manual curation. In addition, it would be valuable for biomedical researchers if such approaches would require little method development and instead draw from cutting-edge research such as image classification via deep neural networks. Here, such an approach is outlined that first re-formats protein structures as 2D color images and then applies off-the-shelf neural networks for image classification. It is shown that such neural networks can be trained to effectively encode the CATH protein classification database and that feature vectors extracted from such networks, once trained, can be transferred to a completely new task that is likely to benefit from molecular protein information, namely that of small molecule binding.

Statistical Classifications as the Basis for Metadata in Scientific Research

It has been determined that international statistical classifications play the role of standard classifications in one or several statistical areas. The role of the central authority for coordinating work on all statistical classifications is played by the Group of Experts on International Statistical Classifications, created to improve cooperation in the field of improving international classifications, ensuring harmonization and convergence betwe en classifications in the family classes of the International Statistical Classes. It has been determined that the reference structure of information objects that allows universal description of the definition, management and use of data and metadata in scientific research is the GSIM (Generic Statistical Information Model). It has been determined that international statistical classifications play the role of standard classifications in one or several statistical areas. The role of the central authority for coordinating work on all statistical classifications is played by the Group of Experts on International Statistical Classifications, created to improve cooperation in the field of improving international classifications, ensuring harmonization and convergence between classifications in the family classes of the International Statistical Classes. It has been determined that the reference structure of information objects that allows universal description of the definition, management and use of data and metadata in scientific research is the GSIM (Generic Statistical Information Model). According to the terminology of the Model of Statistical Classifications, which includes GSIM, the cases of using the term «classification» in statistics are considered. It was specified that the classification can be linear or have a hierarchical structure. The principles of statistical classification, which must be observed during its construction, include: mutually exclusive, completeness, statistical expediency are provided. The GSIM structural chart of the Statistical Classification Model is given, which provides a conceptual basis for the development of the classification database. The GSIM model of statistical classifications defines a concept in a two-level structure of object types and attributes. It is indicated that on the first level, it defines the main types of objects in the classification database, and at the second level, it lists the attributes associated with each type of object. The types of statistical classifications by the level of distribution are determined, among them: reference, related and derivative classifications. Types of variants of statistical classifications: expansion, aggregation, regrouping are considered. It is noted that a particular version may include elements from more than one of these variants. The main directions of activity of the Group of Experts on international statistical classifications at the present stage are indicated. It has been determined that the main sources of information on international classifications are the RAMON server and the United Nations website.

Confusion Graph: Detecting Confusion Communities in Large Scale Image Classification

For deep CNN-based image classification models, we observe that confusions between classes with high visual similarity are much stronger than those where classes are visually dissimilar. With these unbalanced confusions, classes can be organized in communities, which is similar to cliques of people in the social network. Based on this, we propose a graph-based tool named "confusion graph" to quantify these confusions and further reveal the community structure inside the database. With this community structure, we can diagnose the model's weaknesses and improve the classification accuracy using specialized expert sub-nets, which is comparable to other state-of-the-art techniques. Utilizing this community information, we can also employ pre-trained models to automatically identify mislabeled images in the large scale database. With our method, researchers just need to manually check approximate 3% of the ILSVRC2012 classification database to locate almost all mislabeled samples.