arXiv subjects

Engin Erzin

Publications and source records attributed to Engin Erzin.

14 recordsLinked to original sources

arXiv2025

Learning Annotation Consensus for Continuous Emotion Recognition

In affective computing, datasets often contain multiple annotations from different annotators, which may lack full agreement. Typically, these annotations are merged into a single gold standard label, potentially losing valuable inter-rater variability. We propose a multi-annotator training approach for continuous emotion recognition (CER) that seeks a consensus across all annotators rather than relying on a single reference label. Our method employs a consensus network to aggregate annotations into a unified representation, guiding the main arousal-valence predictor to better reflect collective inputs. Tested on the RECOLA and COGNIMUSE datasets, our approach outperforms traditional methods that unify annotations into a single label. This underscores the benefits of fully leveraging multi-annotator data in emotion recognition and highlights its applicability across various fields where annotations are abundant yet inconsistent.

cs.HC↗

arXiv2024

Efficient and Safe Contact-rich pHRI via Subtask Detection and Motion Estimation using Deep Learning

This paper proposes an adaptive admittance controller for improving efficiency and safety in physical human-robot interaction (pHRI) tasks in small-batch manufacturing that involve contact with stiff environments, such as drilling, polishing, cutting, etc. We aim to minimize human effort and task completion time while maximizing precision and stability during the contact of the machine tool attached to the robot's end-effector with the workpiece. To this end, a two-layered learning-based human intention recognition mechanism is proposed, utilizing only the kinematic and kinetic data from the robot and two force sensors. A ``subtask detector" recognizes the human intent by estimating which phase of the task is being performed, e.g., \textit{Idle}, \textit{Tool-Attachment}, \textit{Driving}, and \textit{Contact}. Simultaneously, a ``motion estimator" continuously quantifies intent more precisely during the \textit{Driving} to predict when \textit{Contact} will begin. The controller is adapted online according to the subtask while allowing early adaptation before the \textit{Contact} to maximize precision and safety and prevent potential instabilities. Three sets of pHRI experiments were performed with multiple subjects under various conditions. Spring compression experiments were performed in virtual environments to train the data-driven models and validate the proposed adaptive system, and drilling experiments were performed in the physical world to test the proposed methods' efficacy in real-life scenarios. Experimental results show subtask classification accuracy of 84\% and motion estimation R\textsuperscript{2} score of 0.96. Furthermore, 57\% lower human effort was achieved during \textit{Driving} as well as 53\% lower oscillation amplitude at \textit{Contact} as a result of the proposed system.

cs.RO↗

arXiv2024

Cluster-to-Predict Affect Contours from Speech

Continuous emotion recognition (CER) aims to track the dynamic changes in a person's emotional state over time. This paper proposes a novel approach to translating CER into a prediction problem of dynamic affect-contour clusters from speech, where the affect-contour is defined as the contour of annotated affect attributes in a temporal window. Our approach defines a cluster-to-predict (C2P) framework that learns affect-contour clusters, which are predicted from speech with higher precision. To achieve this, C2P runs an unsupervised iterative optimization process to learn affect-contour clusters by minimizing both clustering loss and speech-driven affect-contour prediction loss. Our objective findings demonstrate the value of speech-driven clustering for both arousal and valence attributes. Experiments conducted on the RECOLA dataset yielded promising classification results, with F1 scores of 0.84 for arousal and 0.75 for valence in our four-class speech-driven affect-contour prediction model.

eess.AS↗

arXiv2022

Role of Audio in Audio-Visual Video Summarization

Video summarization attracts attention for efficient video representation, retrieval, and browsing to ease volume and traffic surge problems. Although video summarization mostly uses the visual channel for compaction, the benefits of audio-visual modeling appeared in recent literature. The information coming from the audio channel can be a result of audio-visual correlation in the video content. In this study, we propose a new audio-visual video summarization framework integrating four ways of audio-visual information fusion with GRU-based and attention-based networks. Furthermore, we investigate a new explainability methodology using audio-visual canonical correlation analysis (CCA) to better understand and explain the role of audio in the video summarization task. Experimental evaluations on the TVSum dataset attain F1 score and Kendall-tau score improvements for the audio-visual video summarization. Furthermore, splitting video content on TVSum and COGNIMUSE datasets based on audio-visual CCA as positively and negatively correlated videos yields a strong performance improvement over the positively correlated videos for audio-only and audio-visual video summarization.

cs.CV↗

arXiv2022

Experiments on Turkish ASR with Self-Supervised Speech Representation Learning

While the Turkish language is listed among low-resource languages, literature on Turkish automatic speech recognition (ASR) is relatively old. In this report, we present our findings on Turkish ASR with speech representation learning using HUBERT. We investigate pre-training HUBERT for Turkish with large-scale data curated from online resources. We pre-train our model using 6,500 hours of speech data from YouTube. The results show that the models are not ready for commercial use since they are not robust against disturbances that typically occur in real-world settings such as variations in accents, slang, background noise and interference. We analyze typical errors and the limitations of the models for use in commercial settings.

cs.CL↗

arXiv2022

Audience Response Prediction from Textual Context

Humans' perception system closely monitors audio-visual cues during multiparty interactions to react timely and naturally. Learning to predict timing and type of reaction responses during human-human interactions may help us to enrich human-computer interaction applications. In this paper we consider a presenter-audience setting and define an audience response prediction task from the presenter's textual speech. The task is formulated as a binary classification problem as occurrence and absence of response after the presenter's textual speech. We use the BERT model as our classifier and investigate models with different textual contexts under causal and non-causal prediction settings. While the non-causal textual context, one sentence preceding and one sentence following the response event, can hugely improve the accuracy of predictions, we showed that longer textual contexts with causal settings attain UAR and $F1$-Score improvements matching and exceeding the non-causal textual context performance within the experimental evaluations on the OPUS and TED datasets.

cs.HC↗

arXiv2021

Affective Burst Detection from Speech using Kernel-fusion Dilated Convolutional Neural Networks

As speech-interfaces are getting richer and widespread, speech emotion recognition promises more attractive applications. In the continuous emotion recognition (CER) problem, tracking changes across affective states is an important and desired capability. Although CER studies widely use correlation metrics in evaluations, these metrics do not always capture all the high-intensity changes in the affective domain. In this paper, we define a novel affective burst detection problem to accurately capture high-intensity changes of the affective attributes. For this problem, we formulate a two-class classification approach to isolate affective burst regions over the affective state contour. The proposed classifier is a kernel-fusion dilated convolutional neural network (KFDCNN) architecture driven by speech spectral features to segment the affective attribute contour into idle and burst sections. Experimental evaluations are performed on the RECOLA and CreativeIT datasets. The proposed KFDCNN is observed to outperform baseline feedforward neural networks on both datasets.

cs.SD↗

arXiv2021

Use of Affective Visual Information for Summarization of Human-Centric Videos

Increasing volume of user-generated human-centric video content and their applications, such as video retrieval and browsing, require compact representations that are addressed by the video summarization literature. Current supervised studies formulate video summarization as a sequence-to-sequence learning problem and the existing solutions often neglect the surge of human-centric view, which inherently contains affective content. In this study, we investigate the affective-information enriched supervised video summarization task for human-centric videos. First, we train a visual input-driven state-of-the-art continuous emotion recognition model (CER-NET) on the RECOLA dataset to estimate emotional attributes. Then, we integrate the estimated emotional attributes and the high-level representations from the CER-NET with the visual information to define the proposed affective video summarization architectures (AVSUM). In addition, we investigate the use of attention to improve the AVSUM architectures and propose two new architectures based on temporal attention (TA-AVSUM) and spatial attention (SA-AVSUM). We conduct video summarization experiments on the TvSum database. The proposed AVSUM-GRU architecture with an early fusion of high level GRU embeddings and the temporal attention based TA-AVSUM architecture attain competitive video summarization performances by bringing strong performance improvements for the human-centric videos compared to the state-of-the-art in terms of F-score and self-defined face recall metrics.

cs.CV↗

arXiv2020

AffectON: Incorporating Affect Into Dialog Generation

Due to its expressivity, natural language is paramount for explicit and implicit affective state communication among humans. The same linguistic inquiry (e.g., How are you?) might induce responses with different affects depending on the affective state of the conversational partner(s) and the context of the conversation. Yet, most dialog systems do not consider affect as constitutive aspect of response generation. In this paper, we introduce AffectON, an approach for generating affective responses during inference. For generating language in a targeted affect, our approach leverages a probabilistic language model and an affective space. AffectON is language model agnostic, since it can work with probabilities generated by any language model (e.g., sequence-to-sequence models, neural language models, n-grams). Hence, it can be employed for both affective dialog and affective language generation. We experimented with affective dialog generation and evaluated the generated text objectively and subjectively. For the subjective part of the evaluation, we designed a custom user interface for rating and provided recommendations for the design of such interfaces. The results, both subjective and objective demonstrate that our approach is successful in pulling the generated language toward the targeted affect, with little sacrifice in syntactic coherence.

cs.CL↗

arXiv2020

Multimodal Continuous Emotion Recognition using Deep Multi-Task Learning with Correlation Loss

In this study, we focus on continuous emotion recognition using body motion and speech signals to estimate Activation, Valence, and Dominance (AVD) attributes. Semi-End-To-End network architecture is proposed where both extracted features and raw signals are fed, and this network is trained using multi-task learning (MTL) rather than the state-of-the-art single task learning (STL). Furthermore, correlation losses, Concordance Correlation Coefficient (CCC) and Pearson Correlation Coefficient (PCC), are used as an optimization objective during the training. Experiments are conducted on CreativeIT and RECOLA database, and evaluations are performed using the CCC metric. To highlight the effect of MTL, correlation losses and multi-modality, we respectively compare the performance of MTL against STL, CCC loss against root mean square error (MSE) loss and, PCC loss, multi-modality against single modality. We observe significant performance improvements with MTL training over STL, especially for estimation of the valence. Furthermore, the CCC loss achieves more than 7% CCC improvements on CreativeIT, and 13% improvements on RECOLA against MSE loss.

cs.HC↗

arXiv2019

Emotion Dependent Facial Animation from Affective Speech

In human-to-computer interaction, facial animation in synchrony with affective speech can deliver more naturalistic conversational agents. In this paper, we present a two-stage deep learning approach for affective speech driven facial shape animation. In the first stage, we classify affective speech into seven emotion categories. In the second stage, we train separate deep estimators within each emotion category to synthesize facial shape from the affective speech. Objective and subjective evaluations are performed over the SAVEE dataset. The proposed emotion dependent facial shape model performs better in terms of the Mean Squared Error (MSE) loss and in generating the landmark animations, as compared to training a universal model regardless of the emotion.

eess.AS↗

arXiv2019

Batch Recurrent Q-Learning for Backchannel Generation Towards Engaging Agents

The ability to generate appropriate verbal and non-verbal backchannels by an agent during human-robot interaction greatly enhances the interaction experience. Backchannels are particularly important in applications like tutoring and counseling, which require constant attention and engagement of the user. We present here a method for training a robot for backchannel generation during a human-robot interaction within the reinforcement learning (RL) framework, with the goal of maintaining high engagement level. Since online learning by interaction with a human is highly time-consuming and impractical, we take advantage of the recorded human-to-human dataset and approach our problem as a batch reinforcement learning problem. The dataset is utilized as a batch data acquired by some behavior policy. We perform experiments with laughs as a backchannel and train an agent with value-based techniques. In particular, we demonstrate the effectiveness of recurrent layers in the approximate value function for this problem, that boosts the performance in partially observable environments. With off-policy policy evaluation, it is shown that the RL agents are expected to produce more engagement than an agent trained from imitation learning.

cs.AI↗

arXiv2019

Speech Driven Backchannel Generation using Deep Q-Network for Enhancing Engagement in Human-Robot Interaction

We present a novel method for training a social robot to generate backchannels during human-robot interaction. We address the problem within an off-policy reinforcement learning framework, and show how a robot may learn to produce non-verbal backchannels like laughs, when trained to maximize the engagement and attention of the user. A major contribution of this work is the formulation of the problem as a Markov decision process (MDP) with states defined by the speech activity of the user and rewards generated by quantified engagement levels. The problem that we address falls into the class of applications where unlimited interaction with the environment is not possible (our environment being a human) because it may be time-consuming, costly, impracticable or even dangerous in case a bad policy is executed. Therefore, we introduce deep Q-network (DQN) in a batch reinforcement learning framework, where an optimal policy is learned from a batch data collected using a more controlled policy. We suggest the use of human-to-human dyadic interaction datasets as a batch of trajectories to train an agent for engaging interactions. Our experiments demonstrate the potential of our method to train a robot for engaging behaviors in an offline manner.

cs.AI↗

arXiv2009

Inter Genre Similarity Modelling For Automatic Music Genre Classification

Music genre classification is an essential tool for music information retrieval systems and it has been finding critical applications in various media platforms. Two important problems of the automatic music genre classification are feature extraction and classifier design. This paper investigates inter-genre similarity modelling (IGS) to improve the performance of automatic music genre classification. Inter-genre similarity information is extracted over the mis-classified feature population. Once the inter-genre similarity is modelled, elimination of the inter-genre similarity reduces the inter-genre confusion and improves the identification rates. Inter-genre similarity modelling is further improved with iterative IGS modelling(IIGS) and score modelling for IGS elimination(SMIGS). Experimental results with promising classification improvements are provided.

cs.SD↗