Cross-Corpus Speech Emotion Recognition Based on Transfer Learning and Multi-Loss Dynamic Adjustment

Table 1

A brief summary of related work.

References	Year	Methods	Features	Corpus
Zong et al. [13]	2016	Least squares regression	INTERSPEECH 2009	Berlin, AFEW 4.0, eNTERFACE
Liu et al. [14]	2018	Feature selection + SVM	INTERSPEECH 2009	Berlin, AFEW 4.0, eNTERFACE
Luo et al. [15]	2019	NMF + MMD	Segmental features	Berlin, CASIA, eNTERFACE, Estonian
Song [16]	2019	TLSL	INTERSPEECH 2010	Berlin, FAU-AIBO, eNTERFACE
Zhang et al. [17]	2020	TSDSL	INTERSPEECH 2010	Berlin, BAUM-1a, eNTERFACE
Zhang et al. [18]	2021	JDAR	INTERSPEECH 2010	Berlin, CASIA, eNTERFACE
Zehra et al. [19]	2021	Ensemble learning	Spectral and prosodic	SAVEE, UrduRDU, EMO-DB, EMOVO
Latif et al. [28]	2018	DBNs	eGeMAPS feature set	FAU-AIBO, SAVEE IEMOCAP, EMO-DB, EMOVO
Zhang et al. [29]	2019	Deep metric learning	Log Mel-frequencyfilter-bank energy	IEMOCAP, MSP-improv
Ahn et al. [30]	2021	Few-shot learning	INTERSPEECH 2010	IEMOCAP, CREMA-D, MSP-IMPROV,Berlin, Korean multimodal emotion dataset
Chang et al. [31]	2021	Adversarial learning	INTERSPEECH 2010	IEMOCAP, MSP-improv, MSP-PODCAST
Sneha et al. [32]	2022	VAE with KL annealing	eGeMAPS feature set	IEMOCAP, SAVEE, Berlin, CaFE, URDU, AESD