Attention Networks for Modeling Behaviors in Addiction Counseling

Gibson, James; Can, Doğan; Georgiou, Panayiotis G.; Atkins, David C.; Narayanan, Shrikanth

doi:10.21437/interspeech.2017-218

Cited by 20 publications

(21 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several computational models have been proposed for predicting MISC behavioral codes at the utterance level [2,3,9]. Researchers have addressed this problem by using variety of features, such as word n-grams and linguistic features [1] and recurrent neural networks (RNNs) with word embedding features [3,11]. Methods using RNNs have shown superior performance to other models (e.g., MaxEnt) for utterance level behavioral code prediction [3].…”

Section: Related Workmentioning

confidence: 99%

“…Self-attention mechanisms, which enable models to attend to particular words based on input for predicting output classes, have been used widely in natual language processing [12,13,14] and speech processing [15]. Recently [11] extended the work from [3] by using a self-attention mechanism for predicting utterance level MISC codes. They show how attention can improve the interpretability and help in better understanding the decisions made by the model.…”

Section: Related Workmentioning

confidence: 99%

“…We show that prosodic information can assist lexical information in making a multimodal prediction. Our multimodal architecture is largely inspired from [11] and [3]. Thus, we use Bidirectional long short term memory (LSTM) networks with a self-attention mechanism for predicting MISC Codes at the level of utterances using multimodal information i.e prosodic and lexical features.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Using Prosodic and Lexical Information for Learning Utterance-level Behaviors in Psychotherapy

et al. 2018

Self Cite

View full text Add to dashboard Cite

In this paper, we present an approach for predicting utterance level behaviors in psychotherapy sessions using both speech and lexical features. We train long short term memory (LSTM) networks with an attention mechanism using words, both manually and automatically transcribed, and prosodic features, at the word level, to predict the annotated behaviors. We demonstrate that prosodic features provide discriminative information relevant to the behavior task and show that they improve prediction when fused with automatically derived lexical features. Additionally, we investigate the weights of the attention mechanism to determine words and prosodic patterns which are of importance to the behavior prediction task.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Using Prosodic and Lexical Information for Learning Utterance-level Behaviors in Psychotherapy

et al. 2018

Self Cite

View full text Add to dashboard Cite

show abstract

“…Although most of the research efforts focus on the therapist's language and speech, it has been shown that examining patient's language can be beneficial for specific behavior cues [9]. More recently, text-only approaches to assess MI have been possible thanks to deep learning models [10][11][12][13].…”

Section: Introductionmentioning

confidence: 99%

Language Features for Automated Evaluation of Cognitive Behavior Psychotherapy Sessions

et al. 2018

Self Cite

View full text Add to dashboard Cite

Cognitive Behavior Therapy (CBT) is a psychotherapy treatment that uses cognitive change strategies to address mental health problems. Quality assessment of a CBT session is traditionally addressed by human raters who evaluate recorded sessions along specific behavioral codes, a cost prohibitive and time consuming method. In this work we examine how linguistic features can be effectively used to develop an automatic competency rating tool for CBT. We explore both standard, widelyused lexical features and domain-specific ones, adapting methods which have been successfully used in similar psychotherapy session coding tasks. Experiments are conducted on manual transcripts of CBT sessions and on automatically derived ones, thus introducing an end-to-end approach. Our results suggest that a real-world system could be developed to automatically evaluate CBT sessions to assist training, supervision, or quality assurance of services.

show abstract

“…Several methods for automatic assessment of treatment effectiveness have been proposed. Most of them rely on audio and linguistic cues (e.g., empathy and behavioral codes in therapy [28,29]). Other methods explored the use of unsupervised topic modelling techniques as a higher-level measure of content, specifically their relation to mental health outcomes [30,31].…”

Section: Previous Workmentioning

confidence: 99%

Identifying Therapist and Client Personae for Therapeutic Alliance Estimation

et al. 2019

Self Cite

View full text Add to dashboard Cite

Psychotherapy, from a narrative perspective, is the process in which a client relates an ongoing life-story to a therapist. In each session, a client will recount events from their life, some of which stand out as more significant than others. These significant stories can ultimately shape one's identity. In this work we study these narratives in the context of therapeutic alliance-a self-reported measure on the perception of a shared bond between client and therapist. We propose that alliance can be predicted from the interactions between certain types of clients with types of therapists. To validate this method, we obtained 1235 transcribed sessions with client-reported alliance to train an unsupervised approach to discover groups of therapists and clients based on common types of narrative characters, or personae. We measure the strength of the relation between personae and alliance in two experiments. Our results show that (1) alliance can be explained by the interactions between the discovered character types, and (2) models trained on therapist and client personae achieve significant performance gains compared to competitive supervised baselines. Finally, exploratory analysis reveals important character traits that lead to an improved perception of alliance.

show abstract

Attention Networks for Modeling Behaviors in Addiction Counseling

Cited by 20 publications

References 16 publications

Using Prosodic and Lexical Information for Learning Utterance-level Behaviors in Psychotherapy

Using Prosodic and Lexical Information for Learning Utterance-level Behaviors in Psychotherapy

Language Features for Automated Evaluation of Cognitive Behavior Psychotherapy Sessions

Identifying Therapist and Client Personae for Therapeutic Alliance Estimation

Contact Info

Product

Resources

About