Does Tail Label Help for Large-Scale Multi-Label Learning?

08:00 EDT 10th September 2019 | BioPortfolio

Summary of "Does Tail Label Help for Large-Scale Multi-Label Learning?"

Large-scale multi-label learning (LMLL) annotates relevant labels for unseen data from a huge number of candidate labels. It is perceived that labels exhibit a long tail distribution in which a significant number of labels are tail labels. Most previous studies consider that the performance would benefit from incorporating tail labels. Nonetheless, it is not quantified how tail labels impact the performance. In this article, we disclose that whatever labels are randomly missing or misclassified, the impact of labels on commonly used LMLL evaluation metrics (Propensity Score Precision (PSP)@k and Propensity Score nDCG (PSnDCG)@k) is directly related to the product of the label weights and the label frequencies. In particular, when labels share equal weights, tail labels impact much less than common labels due to the scarcity of relevant examples. Based on such observation, we propose to develop low-complexity LMLL methods with the goal of facilitating fast prediction time and compact model size by restraining less performance-influential labels. With the consideration that discarding labels may cause the loss of predictive capability, we further propose to preserve dominant model parameters for the less performance-influential labels. Experiments clearly justify that both the prediction time and the model size are significantly reduced without sacrificing much predictive performance.


Journal Details

This article was published in the following journal.

Name: IEEE transactions on neural networks and learning systems
ISSN: 2162-2388


DeepDyve research library

PubMed Articles [14067 Associated PubMed Articles listed on BioPortfolio]

In silico prediction of endocrine disrupting chemicals using single-label and multi-label models.

Endocrine disruption (ED) has become a serious public health issue and also poses a significant threat to the ecosystem. Due to complex mechanisms of ED, traditional in silico models focusing on only ...

Label-specific feature selection and two-level label recovery for multi-label classification with missing labels.

In multi-label learning, each instance is assigned by several nonexclusive labels. However, these labels are often incomplete, resulting in unsatisfactory performance in label related applications. We...

Call for Action: Benzodiazepine Prescription Prevalence Analysis Shows Off-Label Prescription in One in Eleven Citizens.

Benzodiazepines have a good safety profile. Nevertheless, off-label use of these drugs may increase the risk of falls, dependence, and memory loss. Off-label prescription use is not highly prevalent. ...

Quantitative Phosphoproteomic Using Titanium Dioxide Micro-Columns and Label-Free Quantitation.

Phosphorylation events are important during cellular function. Analysis of phosphorylation in complex samples has been extensively studied using large-scale phosphopeptide enrichment methods. Quantita...

Deep Integration: A Multi-label Architecture for Road Scene Recognition.

Deep convolutional neural networks have been applied by automobile industries, internet giants and academic institutes to boost autonomous driving technologies, while progress has been witnessed in en...

Clinical Trials [14632 Associated Clinical Trials listed on BioPortfolio]

The Influence of Multi-domain Cognitive Training on Large-scale Structural and Functional Brain Networks in MCI

The purpose of this study is to integrate advanced computational techniques and multimodal neuroimaging methods to examine the potential effects of long-term, multi-domain, online, compute...

Feasibility of Olanzapine at REduced doSe in hIGHly Emetogenic chemoTherapy

Olanzapine is frequently used off-label as an adjunct antiemetic in clinical oncology settings. North American oncology guidelines recommend it as salvage therapy and as add-on to the stan...

Wolverine: PNT2258 for Treatment of Patients With r/r DLBCL

This study is a multi-center, nonrandomized, open label, phase II investigation of PNT2258 to characterize anti-tumor activity and collect safety data on patients with relapsed or refracto...

Laparoscopic vs Open Pancreatectomy for Body and Tail Pancreatic Cancer

Open distal pancreatectomy (ODP) has been commonly employed for the treatment of a variety of cancers in body and tail of pancreas. Although many general surgical procedures have been incr...

Open Label Multi-Site Study of Safety and Effects of MDMA-assisted Psychotherapy for Treatment of PTSD

This multi-site, open-label, Phase 2, lead-in study assesses the safety and effect of 3,4-methylenedioxymethamphetamine (MDMA)-assisted psychotherapy in participants diagnosed with at leas...

Medical and Biotech [MESH] Definitions

The practice of prescribing or using a drug outside the scope of the drug's official approved label as designated by a regulatory agency concerning the treatment of a particular disease or condition.

A photoactivable URIDINE analog that is used as an affinity label.

Cyclic N-oxide radical functioning as a spin label and radiation-sensitizing agent.

Methods used to label and follow the course of NEURAL PATHWAYS by AXONAL TRANSPORT of injected NEURONAL TRACT-TRACERS.

The act of ligating UBIQUITINS to PROTEINS to form ubiquitin-protein ligase complexes to label proteins for transport to the PROTEASOME ENDOPEPTIDASE COMPLEX where proteolysis occurs.

Quick Search

DeepDyve research library

Searches Linking to this Article