Parabel

Prabhu, Yashoteja; Kag, Anil; Harsola, Shrutendra; Agrawal, Rahul; Varma, Manik

doi:10.1145/3178876.3185998

Cited by 138 publications

(28 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In contrast, the algorithmic complexity of kNN scales logarithmically with the number of training points used. For this reason, tree-based methods are common in other Extreme Classification applications [9], where similarly massive numbers of categories are under consideration. Unfortunately, the resulting faster runtimes come at a significant cost; kNN often gives far worse results in practice than a more rigorous Bayesian approach.…”

Section: Methodsmentioning

confidence: 99%

“…For example, in the human proteome, there are about 20,000 proteins, which when processed with an amino-acid specific protease such as trypsin can correspond to hundreds of thousands or even millions of distinct peptides, each of which can potentially vary due to post-translational modifications or experiment-specific processing. This puts fluorosequencing data analysis squarely in the realm of Extreme Classification problems, which are known to be challenging to handle in practice [9].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Amino acid sequence assignment from single molecule peptide sequencing data using a two-stage classifier

Smith

Simpson²,

Marcotte

2022

Preprint

View full text Add to dashboard Cite

We present a machine learning-based interpretive framework (whatprot) for analyzing single molecule protein sequencing data produced by fluorosequencing, a recently developed proteomics technology that determines sparse amino acid sequences for many individual peptide molecules in a highly parallelized fashion [1] [2]. Whatprot uses Hidden Markov Models (HMMs) to represent the states of each peptide undergoing the various chemical processes during fluorosequencing, and applies these in a Bayesian classifier, in combination with pre-filtering by a k-Nearest Neighbors (kNN) classifier trained on large volumes of simulated fluorosequencing data. We have found that by combining the HMM based Bayesian classifier with the kNN pre-filter, we are able to retain the benefits of both, achieving both tractable runtimes and acceptable precision and recall for identifying peptides and their parent proteins from complex mixtures, outperforming the capabilities of either classifier on its own. Whatprot's hybrid kNN-HMM approach enables the efficient interpretation of fluorosequencing data using a full proteome reference database and should now also enable improved sequencing error rate estimates.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Amino acid sequence assignment from single molecule peptide sequencing data using a two-stage classifier

Smith

Simpson²,

Marcotte

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…This is a basic approach, focusing on improving model efficiency, but requiring an extremely large amount of computational complexity, typically represented by DiSMEC [5] and PPDSparse [6]; 2) The main intention of Tree-based methods is to overcome the high computational complexity of the One-vs-all method. It is based on the idea of performing hierarchical classification, where the features of the labels can be implemented in clusters, represented by Parabel [7]; 3) Embedding-based method approaches that perform high-and low-dimensional spatial transformations through label compression and decompression, sacrificing a certain amount of information in exchange for a reduction in computational complexity, represented by AnnexML [8]; 4) Deep learning method is the current mainstream direction. Since 2017, XML-CNN has become the first successful deep learning method in the XMC domain and introduced the important concept of bottleneck layer.…”

Section: Related Workmentioning

confidence: 99%

“…We compare the most representative methods as our baselines including: DiSMEC [5] is a largescale distributed framework for learning 1-vs-All approaches. Parabel [7], a method using PLTs, can be considered as taking the advantage of both tree-based and 1-vs-All methods. Bonsai [16], another method using PLTs, is similar to Parabel.…”

Section: Baselinesmentioning

confidence: 99%

CRAT-XML: Contrastive Representation Adversarial Training for Extremely Multi-Label Text Classification

Zhu

Huang

Mao

2023

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

Extreme multi-label text classification is a very important and challenging problem in this age of widespread internet access, with a wide range of application scenarios, such as web tagging, legal document annotation, commodity classification, etc. Most of the existing state-of-the-art methods are based on deep learning, however, there are still two problems: 1) the vector representation generated by most pre-trained models suffers from anisotropy and uneven distribution, which has a significant impact on the XMC task. 2) existing models are large in size and use many models for integration. Some of them even took hundreds of hours to train. This seriously affects the efficiency of the experiments. Therefore, in this paper, we propose CRAT-XML, which uses contrast adversarial learning to optimize text representation and enhance the acquisition of dependency relations between text and labels, thus reducing the need for integration at the representation level and achieving relatively high accuracy under low-resource and low-time conditions. Experimental results demonstrate that our model achieves SOTA results on a single model, while achieving a large reduction in training time and model size.

show abstract

“…Many machine learning based systems for automated subject indexing have been developed since the 1990s, when the approach became the dominant paradigm for automated subject indexing (Sebastiani, 2002). Some recent examples for which an implementation is available include Magpie (Berger, 2015;Kim, 2014), FastXML (Prabhu & Varma, 2014), PD-Sparse (Yen, Huang, Zhong, Ravikumar, & Dhillon, 2016), fastText (Joulin, Grave, Bojanowski, & Mikolov, 2017), Quadflor (Galke, Mai, Schelten, Brunsch, & Scherp, 2017), AnnexML (Tagami, 2017) and Parabel (Prabhu, Kag, Harsola, Agrawal, & Varma, 2018).…”

Section: Machine Learning Approachesmentioning

confidence: 99%

Annif: DIY automated subject indexing using multiple algorithms

Suominen¹

2019

LIBER

View full text Add to dashboard Cite

Manually indexing documents for subject-based access is a labour-intensive process. We propose using metadata gathered from bibliographic databases to train algorithms that assist librarians in that work. We have developed Annif, an open source tool and microservice for automated subject indexing. After training it with a subject vocabulary and existing metadata, Annif can be used to assign subject headings for new documents. We have tested Annif with different document collections including scientific papers, old scanned books and contemporary e-books, Q&A pairs from an "ask a librarian" service, Finnish Wikipedia, and the archives of a local newspaper. The results of analysing scientific papers and current books have been reassuring, while other types of documents have proved to be more challenging. The current version is based on a combination of existing natural language processing and machine learning tools. By combining multiple approaches and existing open source algorithms, Annif can build on the strengths of individual algorithms and adapt to different settings. With Annif, we expect to improve subject indexing and classification processes especially for electronic documents as well as collections that otherwise would not be indexed at all.

show abstract

Parabel

Cited by 138 publications

References 18 publications

Amino acid sequence assignment from single molecule peptide sequencing data using a two-stage classifier

Amino acid sequence assignment from single molecule peptide sequencing data using a two-stage classifier

CRAT-XML: Contrastive Representation Adversarial Training for Extremely Multi-Label Text Classification

Annif: DIY automated subject indexing using multiple algorithms

Contact Info

Product

Resources

About