Emotion Built Stress Relief Recommender Tool Using Machine Learning

doi:10.21203/rs.3.rs-1325981/v1

Nowadays, Society and the operating surroundings positioned an unrivalled diploma of consciousness on individuals. The indicators of pressure are ubiquitous, and its results are various. The mitigating depth of track is settled. It has a one in every of a type connect with our feelings, so may be a totally a success stress the board device. Tuning in to track will have a highly loosening up effect on our brains and bodies, especially moderate, calm vintage fashion track. This form of track can valuably have an effect on our physiological limits, shifting again the beat and heartbeat, slicing down circulatory stress, and lowering the stages of pressure hormones. Nonetheless, it is probably unhelpful if the track from time to time falls quick for the existing feeling of the audience. In addition, there may be no track participant which could select tunes depending on the consumer feeling.

Machine Learning

API

Stress

Image Processing

Face Detection

In the preceding now no longer many years, the extent of people who've stress has raised at a better rate. This is a right away end result of some reasons, for example, obligations, better object value, lousy economy, excessive normal costs, and so on [1]. In 2017, the department of emotional wellbeing gathers information from Telephone management for mental wellbeing publications and gives that the stress of Thai people could in fashionable increment. The amount of calls is in extra of 30,000 calls which might be two times in 2019.

There are several programs available for track participant and stress remedy and the board, however there's no software that could select melodies, Movies, Books depending on the purchaser feeling. [2–3] To address those restrictions, this flexible track participant software that could prescribe melodies depending on the purchaser feeling may be utilized. Initially the software snaps a picture graph of the purchaser. [5] At the factor whilst the software receives the purchaser's face image graph from the flexible camera, It examinations what's the purchaser's gift feeling (dismal, cheerful, livid or impartial). To dissect the purchaser feeling from the face image graph the software makes use of the Google Vision API, making use of which the purchaser image graph .[7]

Utilizing the organized set that is placed away through Google at the Google cloud the sensation is broke down and outstanding from the given face photograph. The Google's Emotion acknowledgments API go back the sensation of the customer depending on the face photograph to the software. After getting the Emotion subtleties from the Google API, Based on the sensation that is diagnosed utilising the face photograph the software offers numerous guidelines to decrease the stress degree to the customer. Based at the customer's gift feeling the software will advise Books, Songs, Movies and so forth. Steps in pre-trained models

This is a conventional instance of Computer Vision which by and large offers with extraction of textual content from a photo. The Vision API accommodates many state-of- the-artwork processes for doing this. This is the project which you executed in the sooner section. With Vision API you may retrieve trendy attributes of a photo, functions which includes dominant colour.

Given a photo or a hard and fast of images, the project is to stumble on the faces found in them. This has numerous big programs like Surveillance Systems.[8] These are a number of the top notch use-instances on which Vision API plays seamlessly, and you may combine any of the above into your programs inside very much less quantity of time. Vision API provides support for a wide range of languages like Go, C#, Java, PHP, Node.js, Python, and Ruby. In the following sections, you may see a way to use Vision API in Python.

Pattern popularity is the technique of spotting styles via way of means of the usage of gadget studying algorithm. Pattern reputation can be defined due to the fact the class of records based totally mostly on understanding already acquired or on statistical facts extracted from patterns and/or their representation. One of the crucial factors of the sample reputation is its utility potential. Pattern reputation includes category and cluster of patterns.

Electroencephalogram (EEG) signal-primarily based totally feeling acknowledgment has pulled in extensive pursuits recently and has been significantly acquired in clinical, emotional registering, and different huge fields. In any case, maximum of the exploration discovered proper now to pay attention at the precision of association even as ignoring the interpretability of feeling movement. Right now, advocate every other interpretable feeling acknowledgment method with the enactment device with the aid of using making use of AI and EEG signals. This paper inventively proposes the enthusiastic actuation bend to reveal the initiation technique of emotions. [9] The calculation first concentrates highlights from EEG flags and institution’s emotions making use of AI methods, wherein diverse portions of an initial are applied to put together the proposed show and survey its impact on feeling acknowledgment results. Second, novel initiation bends of emotions are constructed depending on the association results, and feeling coefficients, i.e., the relationship coefficients and entropy coefficients. The initiation bend can institution emotions in addition to uncover to a constricted diploma the enthusiastic actuation system. At last, a weight coefficient is received from the 2 coefficients to enhance the exactness of feeling acknowledgment. To approve the proposed technique, assessments were accomplished at the DEAP and SEED dataset. [10–11] The effects bolster the factor that emotions are logically actuated during the examination, and the weighting coefficients depending on the relationship coefficient and the entropy coefficient can thoroughly enhance the EEG-primarily based totally feeling acknowledgment exactness.

The take a look at of missing familiarity with passionate incitement aspect in maximum traditional examinations, we proposed a coefficients-prepare method primarily based totally with appreciate to AI using EEG alerts. This method now no longer simply beat the benchmark calculations as some distance as precision but moreover deciphers the development of feeling initiation.

Initially, we eliminated highlights from EEG alerts [Fig. 1] and grouped emotions using AI methods. We in addition located that the final section of EEG alerts have higher connections with emotions, therefore higher classifier execution may be done if the second one 50% of the initial is applied for preparing[Fig. 2]. Furthermore, in mild of the grouping results, the connection bends and entropy bends of emotions are developed, which to a restricted diploma display the passionate initiation movement. It is located that feeling become constantly enacted. The proposed method has given a quantitative tool to hypothetically make clear passionate actuation system, for example, why the second one 50% of an initial activates higher grouping outcomes. At final, the obtained connection coefficients and entropy coefficients are applied to construct weight coefficients to enhance the association exactness contrasted with present day benchmark calculations. Since the burden coefficient relies upon on the connection coefficient and entropy coefficient, the legitimacy of the burden coefficient likewise suggests the legitimacy of the proposed connection coefficient and entropy coefficient from some other angle, which demonstrates that the proposed speculation of dynamic enactment of emotions is sensible.

Measuring the EEG signals is complicate.

Recognizing the signal range of each emotion is very complex.
This system requires lot of hardware components which makes it has in-compact.
The initial cost for implementing this module is very high.
The mission learning algorithm makes the system as complex and hard to maintain

"A statement to confirm that all methods were carried out in accordance with relevant guidelines and regulations”.

The utility receives facial image of the purchaser from the flexible camera. At that factor the facial image is transferred at the utility for image preparing. This is despatched to the Google Vision cloud for image acknowledgment. The face image is carried out at the VISION API which makes use of the organized informational collections of the Google Vision cloud to understand the sensation of the face image.[Fig. 3] Then the existing feeling of the purchaser is distinguished (Sad, Happy, Neutral or Angry) and got here returned to the utility. The administrator continues up the tunes, movement pictures, Books informational collections with inside the Firebase Database.

Then depending on the diagnosed feeling of the purchaser the utility shows the tunes, books and movies to the purchaser. [Fig. 4] This will help the purchaser with lowering his/her stress and to conquer it.

❑ Image recognition Module

❑ Emotion location (Vision API) Module

❑ Songs and Books dataset Module

❑ User Interface Module

Image recognition Module

The software receives facial photograph of the purchaser from the flexible camera/Gallery. [Fig. 5] At that factor the facial photograph is transferred at the software for photograph preparing. This is despatched to the Google cloud for photograph acknowledgment. The Vision API has modules which might be applied to carry out each Face acknowledgment and simply as Emotion acknowledgment. The face acknowledgment module is applied to get the photograph from the purchaser's device and as soon as its miles transferred at the Google's server it play out the region of face with inside the transferred photograph. The Vision API perceive at the least one human faces along qualities, for example, age, feeling, sexual orientation, posture, grin and facial hair, remembering 27 traveller spots for every face for the photograph.

Emotion detection (Vision API) Module

Among the properties, for example, age, feeling, sexual orientation, posture, grin and facial hair, the trait this is required for the application (i.e. Feeling) is amassed with the useful resource of the usage of using the API call to the Google server. The feelings which are gotten from the API server can be in eight unique varieties of feelings Anger, Contempt, Disgust, Fear, Happiness, Neutral, Sadness, Surprise The eight styles of feelings are consolidated into number one four sorts ( i.e. Dismal, Happy, Angry, Neutral ) to decorate the effectiveness and furthermore to decrease the intricacy. Programming interface will deliver the feeling subtleties once more to the application relying on which the winning feeling of the patron can be acquired.

Training set

Training set DEAP is used is used to assemble a model. It consists of the set of images which may be used to teach the device. Training regulations and algorithms used offer relevant facts on a manner to companion input statistics with output decision. The device is professional thru manner of approach of creating use of these algorithms on the dataset, all the relevant facts is extracted from the statistics and results are obtained. [Fig. 6] Generally, 80% of the statistics of the dataset is taken for training statistics.

Songs, Books dataset & User Interface Module

The tunes, movement pix, Books informational indexes are stored up with inside the Firebase Database. The melodies, movement pix and books are categorized for numerous emotions which can be linked with every web page with inside the utility. Set of tunes, Movies and books are associated with their appropriate motion with inside the utility. The utility consists of separate motion for each feeling and the touchdown web page is applied to collect the photograph from the consumer's machine and moreover the ship the photograph to the Google server. [Fig. 7–8] In the wake of having the sensation subtleties from API [Fig. 10], web page which fits with the intestine feeling is displayed to the consumer which includes the stress comfort guidelines like books, tunes which coordinates the consumer's gift feeling.

Vision API plays out the AI calculations utilizing prepared informational index which makes the framework as straightforward and effective.

The exactness of the feeling acknowledgment utilizing VISION API is 0.92199 (i.e. 92.19%) [Table 1]
This framework doesn't requires some other external durable goods with the exception of the cell phone which makes it has smaller.
The framework joins different feelings into significant feelings (dismal, cheerful, irate and nonpartisan) to give better proficiency.
The framework interface and support is easy to comprehend by the client.

Analysis & Discussion

Fig 9 a, b, c, d, e, f, g and h represents the various emotions of the stress with the various techniques and their accuracy. The results show that the various techniques plays a significant role in the various features of the human stress.

Table1: Depicts the accuracy of various techniques

Technique	Accuracy
Feed-Forward Neural Network	82%
PCA and SVM	92%
Neural network	90%
Convolutional Neural Networks (CNN)	60%
k-Nearest Neighbour Algorithm	84%
Support Vector Machine (SVM)	91%
Relevance Vector Machines (RVM)	90.84%
Vision API	92.19%

The key concept is to present the stress remedy proposals to the customers making use of the existing feeling of the client. It makes use of the sensation acknowledgment API which makes the framework as more and more specific and moreover concurrently much less perplexing. Also the accuracy was estimated and compared with various other techniques and was found to be 92.19%. [Fig. 11] Rather than making use of the cutting-edge framework which follows a thoughts boggling method for making use of the EEG to apprehend the emotions the photograph getting ready APIs may be applied to play out this work. Accomplishing those desires is tough whilst making use of a few distinct strategies for locating the sensation of the human, which calls for top notch degree of dataset which makes it as greater complex. This utility improves the Accuracy, Maintainability of the manner which offers higher solace to the client.

Funding: There is no funding support by any funding agency or by anyone.

Conflicts of interest/Competing interests : There is no conflict of Interest.

Availability of data and material: Available on request

Code availability: Available on request

Authors' contributions: Author has done review and written the paper and submitted.

Conflict of Interest: There is no conflict of Interest.

Data availability statements (DAS): All data generated or analysed during this study are included in this published article.

Chankuptarat, K., Sriwatanaworachai, R. and Chotipant, S., 2019, July. Emotion-Based Music Player. In 2019 5th International Conference on Engineering, Applied Sciences and Technology (ICEAST) (pp. 1–4). IEEE.
Quasim, M.T., Alkhammash, E.H., Khan, M.A. and Hadjouni, M., 2021. Emotion-based music recommendation and classification using machine learning with IoT Framework. Soft Computing, pp.1–12.
Kriakous, S.A., Elliott, K.A., Lamers, C. and Owen, R., 2021. The effectiveness of mindfulness-based stress reduction on the psychological functioning of healthcare professionals: A systematic review. Mindfulness, 12(1), pp.1–28.
Kumar, A., Sharma, K. and Sharma, A., 2021. Hierarchical deep neural network for mental stress state detection using IoT based biomarkers. Pattern Recognition Letters, 145, pp.81–87.
Nassif, A.B., Shahin, I., Hamsa, S., Nemmour, N. and Hirose, K., 2021. CASA-based speaker identification using cascaded GMM-CNN classifier in noisy and emotional talking conditions. Applied Soft Computing, 103, p.107141.
Deldjoo, Y., Schedl, M., Cremonesi, P. and Pasi, G., 2020. Recommender systems leveraging multimedia content. ACM Computing Surveys (CSUR), 53(5), pp.1–38.
Yoo, H. and Chung, K., 2020. Deep learning-based evolutionary recommendation model for heterogeneous big data integration. KSII Transactions on Internet and Information Systems (TIIS), 14(9), pp.3730–3744.
Poletti, S., Razzini, G., Ferrari, R., Ricchieri, M.P., Spedicato, G.A., Pasqualini, A., Buzzega, C., Artioli, F., Petropulacos, K., Luppi, M. and Bandieri, E., 2019. Mindfulness-Based stress reduction in early palliative care for people with metastatic cancer: A mixed-method study. Complementary therapies in medicine, 47, p.102218.
Aljanaki, A., Yang, Y.H. and Soleymani, M., 2014, October. Emotion in Music Task at MediaEval 2014. In MediaEval.
Quazi, M.T., 2012. Human emotion recognition using smart sensors. Palmerston North.
Knight, W.E. and Rickard, N.S., 2001. Relaxing music prevents stress-induced increases in subjective anxiety, systolic blood pressure, and heart rate in healthy males and females. Journal of music therapy, 38(4), pp.254–272.
Song, Y., Dixon, S. and Pearce, M.T., 2012, October. Evaluation of musical features for emotion classification. In ISMIR (pp. 523–528).
Yonck, R., 2020. Heart of the machine: Our future in a world of artificial emotional intelligence. Arcade.
Asteriadis, S., Tzouveli, P., Karpouzis, K. and Kollias, S., 2009. Estimation of behavioral user state based on eye gaze and head pose—application in an e-learning environment. Multimedia Tools and Applications, 41(3), pp.469–493.
Wu, C.H., Tzeng, Y.L., Kuo, B.C. and Tzeng, G.H., 2014. Integration of affective computing techniques and soft computing for developing a human affective recognition system for U-learning systems. International Journal of Mobile Learning and Organisation 1, 8(1), pp.50–66.
Garay, N., Cearreta, I., López, J.M. and Fajardo, I., 2006. Assistive technology and affective mediation. Human Technology: an Interdisciplinary Journal on Humans in ICT Environments.
Russell, J.A., 1980. A circumplex model of affect. Journal of personality and social psychology, 39(6), p.1161.

No competing interests reported.

Emotion Built Stress Relief Recommender Tool Using Machine Learning

Status:

Version 1

Abstract

Figures

Introduction

Detection Of Image Properties

Face Detection

Pattern Matching

Existing Work

Proposed Work

Image recognition Module

Emotion detection (Vision API) Module

Training set

Songs, Books dataset & User Interface Module

Conclusion

Declarations

References

Additional Declarations

Status:

Version 1