Pretrained Models with Adversarial Training for Online Sexism Detection @ SemEval 2023

         Abstract 

Adversarial training can provide neural networks with significantly improved resistance to adversarial attacks, thus improving model robustness. However, a major drawback of many existing adversarial training workflows is the computational cost and extra processing time when using data augmentation techniques. This post explores the application of embedding perturbations via the fast gradient method (FGM) when finetuning large language models (LLMs) to short text classification tasks. This adversarial training approach has been evaluated as part of the first sub-task of SemEval 2023-Task 10, focused on explainable detection of sexism in social networks (EDOS). Empirical results show that adversarially finetuned models with FGM had on average a 25% longer training time and 0.2% higher F1 than their respective baselines. 

Introduction

Social media is ubiquitous in everyday communications and a place where users express freely
their views and thoughts. However, while most users make a fair use of social media platforms, the
presence of detrimental and undesirable content that can be deemed abusive towards other users
or communities has drawn lot of attention lately. The threat this poses to online safety has sparked
different content moderation strategies than the traditionally used for fighting spam or phishing, since
in this case they have to deal with legit users occasionally posting messages using toxic, sexist or
abusive language instead of financially motivated cybercrime.

The use of Natural Language Processing (NLP) to detect, and assess sexist content at scale is
widespread, however this is far from being considered a solved problem. The lack of finegrained 
classification and poor interpretability have been highlighted as current shortcomings for this
task. The Explainable Detection of Online Sexism (EDOS) workshop organized as part of SemEval
2023 aims to bridge the above-mentioned gaps in sexist language detection for the English language.

In addition to the identified challenges there are other nuances common to many other language-
based  moderation systems. Social media publications can be short and subject to interpretation
without enough context. Likewise, the presence of slang, spelling variations and purposely obfuscated
content can lower the confidence of NLP tools (Mosquera and Moreda, 2012) and evade detection
(Mosquera, 2022a). Finally, labeling large training sets can be expensive since it requires a
pool of annotators with expert knowledge (Vidgen and Derczynski, 2021). Therefore, training sets
with only a few thousands of positive examples and high class imbalance are not uncommon, which 
can affect model robustness.

Background

The EDOS challenge provided a training dataset (Kirk et al., 2023) of 14000 examples with
 finegrained  classifications for sexist posts written in English extracted from Gab and Reddit social
 networks. Out of these, there were only 3398 entries labeled as sexist, thus existing a substantial 
 class imbalance for the proposed sub-task A, which was binary classification of sexist content.

There is extensive previous work tackling the detection of sexist content in social media using
pre-trained models, either as an standalone task (Rodríguez-Sánchez et al., 2020; Abburi et al.,
2021) or as a subset of toxic language classification (Park and Fung, 2017; Pitsilis et al., 2018).
Some instances of sexist language can be subtle and difficult to identify in absence of proper context
(Swim et al., 2004), specially when working with short texts such as Tweets or isolated sentences
part of a larger conversation. Recent attempts to improve generalization and robustness in these 
cases usually involve ensembles (Davies et al., 2021), adversarial training (Samory et al., 2020) or 
data augmentation (Butt et al., 2021). However, these techniques are not foolproof, and in addition to
 the extra training cost they can introduce unintended bias (Sen et al., 2022).

Approach

For these reasons, in this post we approach the EDOS binary classification task by automatically
finetuning and ensembling sets of different large pre-trained models. The Fast Gradient Method
(FGM), an intuitive backpropagation based method to generate adversarial samples (Szegedy et al.,
2013), is used during training in order to improve model robustness.

The submission produced by the proposed system obtained a 85.6% F1 and ranked 15th out of
84 competing systems. The main takeaway is that adversarial training can be a viable strategy under
domain (text size) or cost (training time) constraints in comparison with other more resource-intensive
techniques such as text augmentation (Mosquera, 2022b).

Results

	Final Ranking	Team Name	Macro F1        Dev Ranking
	1	PingAnLifeInsurance	0.8746	1
	2	stce	0.8740	2
	2	FiRC-NLP	0.8740	3
	4	PALI	0.8717	4
	5	GZHU / UCAS-IIE	0.8692	8
	6	Zhegu	0.8674	9
	7	ITNLP2023	0.8630	13
	8	YD	0.8629	14
	9	MilaNLP	0.8616	17
	10	NLP-LTU	0.8613	19
	11	niceNLP	0.8609	24
	12	CIC-SDS.KN	0.8607	26
	13	cl-uzh	0.8586	30
	14	Aston NLP	0.8583	31
	15	Alejandro Mosquera	0.8562	34
	16	DCU	0.8559	35
	17	Chride	0.8548	38
	18	SUTNLP	0.8545	40
	19	JUST_ONE	0.8538	41
	19	UIRISC	0.8538	42
	19	Just_Dream	0.8538	43
	22	ABC	0.8537	44
	23	MarSan	0.8518	46
	24	UMUTeam	0.8495	47
	25	A2Z	0.8479	50
	26	AKD	0.8461	51
	27	AutoHome_Xu	0.8452	53
	28	PCJ	0.8449	54
	29	LCT-1	0.8446	56
	30	alexa	0.8444	57
	31	tmn	0.8434	58
	32	DH-FBK	0.8402	61
	33	I2C-HUELVA_G2	0.8396	62
	34	SSS	0.8392	63
	35	Attention	0.8390	64
	36	AdamR	0.8383	65
	37	Team Daunting	0.8380	66
	38	DAY	0.8373	67
	39	KingsmanTrio	0.8366	71
	40	Efrat Luzzon	0.8362	72
	41	DUTIR	0.8352	73
	42	UniBoe's	0.8346	74
	43	HULAT	0.8298	75
	44	Group2-RUG	0.8295	76
	45	SKAM	0.8250	77
	46	SINAI	0.8245	79
	47	HHS	0.8239	80
	48	msharma95	0.8230	81
	49	HausaNLP	0.8228	82
	50	UL & UM6P	0.8225	84
	51	CSECU-DSG	0.8218	85
	52	HITSZ-Q	0.8197	87
	53	MaChAmp	0.8192	88
	54	RRGI	0.8186	89
	55	hhuEDOS	0.8185	90
	56	shefnlp	0.8181	91
	57	tsingriver	0.8180	93
	58	JUAGE	0.8166	94
	59	Lexicools	0.8124	96
	60	PA666	0.8119	97
	61	Brainstormers_msec	0.8073	100
	62	ACSMKRHR	0.8009	102
	63	TEDOS	0.8001	103
	64	Stanford MLab	0.7975	104
	65	AU_NLP	0.7943	105
	66	PoSh	0.7937	106
	67	coco	0.7895	108
	68	PadmaDS	0.7826	109
	69	iREL	0.7815	110
	70	shm2023	0.7758	111
	71	IUST_NLP	0.7750	112
	72	INFOTEC	0.7584	115
	73	PanwarJayant	0.7583	116
	74	LT3	0.7542	117
	75	USMBA_NLP	0.7540	118
	76	CNLP-NITS	0.7478	119
	77	YPMS	0.7350	120
	78	we_have_no_idea	0.7304	121
	79	danch22	0.7184	122
	80	NLP-CogSci	0.6325	127
	81	OPEN SESAME	0.6044	128
	82	Judith Jeyafreeda Andrew	0.5191	129
	83	UTB-NLP	0.5185	130
	84	NLP_CHRISTINE	0.5029	131

              https://github.com/rewire-online/edos/blob/main/leaderboard/leaderboard_task_A.csv

References

Harika Abburi, Shradha Sehgal, Himanshu Maheshwari, and Vasudeva Varma. 2021. Knowledge-based neural framework for sexism detection and classification. In IberLEF@ SEPLN, pages 402–414.

Sabur Butt, Noman Ashraf, Grigori Sidorov, and Alexander Gelbukh. 2021. Sexism identification using bert and data augmentation - exist2021. CEUR Workshop Proceedings, 2943:381–389. Publisher Copyright: © 2021 CEUR-WS. All rights reserved.; Conference date: 21-09-2021.

Lily Davies, Marta Baldracchi, Carlo Alessandro Borella, and Konstantinos Perifanos. 2021. Transformer ensembles for sexism detection.

Hannah Rose Kirk, Wenjie Yin, Bertie Vidgen, and Paul Röttger. 2023. SemEval-2023 Task 10: Explainable Detection of Online Sexism. In Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023). Association for Computational Linguistics.

Takeru Miyato, Andrew M. Dai, and Ian Goodfellow. 2016. Adversarial training methods for semisupervised text classification.

Alejandro Mosquera. 2022a. Alejandro mosquera at politices 2022: Towards robust spanish author profiling and lessons learned from adversarial attacks. In Proceedings of the Iberian Languages Evaluation Forum (IberLEF 2022) co-located with the Conference of the Spanish Society for Natural Language Processing (SEPLN 2022), A Coruña, Spain, September 20, 2022, volume 3202 of CEUR Workshop Proceedings. CEUR-WS.org.

Alejandro Mosquera. 2022b. Amsqr at SemEval-2022 task 4: Towards AutoNLP via meta-learning and adversarial data augmentation for PCL detection. In Proceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022), pages 485–489, Seattle, United States. Association for Computational Linguistics.

Alejandro Mosquera and Paloma Moreda. 2012. Tenor: A lexical normalisation tool for spanish web 2.0 texts. In Text, Speech and Dialogue, pages 535–542, Berlin, Heidelberg. Springer Berlin Heidelberg.

Ji Ho Park and Pascale Fung. 2017. One-step and twostep classification for abusive language detection on twitter.

Georgios K. Pitsilis, Heri Ramampiaro, and Helge Langseth. 2018. Effective hate-speech detection in twitter data using recurrent neural networks. Applied Intelligence, 48(12):4730–4742.

Francisco Rodríguez-Sánchez, Jorge Carrillo-de Albornoz, and Laura Plaza. 2020. Automatic classification of sexism in social networks: An empirical study on twitter data. IEEE Access, 8:219563–219576.

Mattia Samory, Indira Sen, Julian Kohne, Fabian Floeck, and Claudia Wagner. 2020. "call me sexist, but...": Revisiting sexism detection using psychological scales and adversarial samples.

Indira Sen, Mattia Samory, Claudia Wagner, and Isabelle Augenstein. 2022. Counterfactually augmented data and unintended bias: The case of sexism and hate speech detection.

Janet K Swim, Robyn Mallett, and Charles Stangor. 2004. Understanding subtle sexism: Detection anduse of sexist language. Sex roles, 51:117–128.

Christian Szegedy, Wojciech Zaremba, Ilya Sutskever, Joan Bruna, Dumitru Erhan, Ian Goodfellow, and Rob Fergus. 2013. Intriguing properties of neural networks.

Bertie Vidgen and Leon Derczynski. 2021. Directions in abusive language training data, a systematic review: Garbage in, garbage out. PLOS ONE, 15(12):1–32.