Table 7: 

Category combinations explored by studies that combine multiple methods. LT indicates whether long tail domains were evaluated on. Method categories explored include feature augmentation (FA), feature generalization (FG), loss augmentation (LA), parameter initialization (PI), ensembling (EN), pseudo-labeling (PL), pretraining (PT), active learning (AL), instance weighting (IW), and data selection (DS).

StudyMethodLT
Different Coarse Categories 
(Jeong et al., 2009) IW+PL ✔ 
(Hangya et al., 2018) PT+FA ✔ 
(Cer et al., 2018) PT+LA ✔ 
(Dereli and Saraclar, 2019) FA+PT ✔ 
(Ji et al., 2015) FG+IW  
(Huang et al., 2019) PI+PL  
(Li et al., 2012) LA+PL+IW  
(Chan and Ng, 2007) AL+PI+IW  
(Nguyen et al., 2014) PL+EN  
(Yu and Kübler, 2011) PL+IW  
(Scheible and Schütze, 2013) FA+PL+DS  
(Tan and Cheng, 2009) FA+IW  
(Mohit et al., 2012) LA+PL  
(Rai et al., 2010) AL+LA  
(Wu et al., 2017) AL+LA  
 
Same Coarse Categories 
(Lin and Lu, 2018) PA+FA ✔ 
(Zhang et al., 2017) FA+LA ✔ 
(Yan et al., 2020) FA+LA  
(Yang et al., 2017) LA+PL+FA  
(Gong et al., 2016) LA+PI  
 
Same Fine Categories 
(Alam et al., 2018) LA+LA ✔ 
(Lee et al., 2020) PL+PL  
(Kim et al., 2017) LA+LA  
StudyMethodLT
Different Coarse Categories 
(Jeong et al., 2009) IW+PL ✔ 
(Hangya et al., 2018) PT+FA ✔ 
(Cer et al., 2018) PT+LA ✔ 
(Dereli and Saraclar, 2019) FA+PT ✔ 
(Ji et al., 2015) FG+IW  
(Huang et al., 2019) PI+PL  
(Li et al., 2012) LA+PL+IW  
(Chan and Ng, 2007) AL+PI+IW  
(Nguyen et al., 2014) PL+EN  
(Yu and Kübler, 2011) PL+IW  
(Scheible and Schütze, 2013) FA+PL+DS  
(Tan and Cheng, 2009) FA+IW  
(Mohit et al., 2012) LA+PL  
(Rai et al., 2010) AL+LA  
(Wu et al., 2017) AL+LA  
 
Same Coarse Categories 
(Lin and Lu, 2018) PA+FA ✔ 
(Zhang et al., 2017) FA+LA ✔ 
(Yan et al., 2020) FA+LA  
(Yang et al., 2017) LA+PL+FA  
(Gong et al., 2016) LA+PI  
 
Same Fine Categories 
(Alam et al., 2018) LA+LA ✔ 
(Lee et al., 2020) PL+PL  
(Kim et al., 2017) LA+LA  
Close Modal

or Create an Account

Close Modal
Close Modal