Recommendation Model Research Articles

DLRover-RM: Resource Optimization for Deep Recommendation Models Training in the Cloud

Deep learning recommendation models (DLRM) rely on large embedding tables to manage categorical sparse features. Expanding such embedding tables can significantly enhance model performance, but at the cost of increased GPU/CPU/memory usage. Meanwhile, tech companies have built extensive cloud-based services to accelerate training DLRM models at scale. In this paper, we conduct a deep investigation of the DLRM training platforms at AntGroup and reveal two critical challenges: low resource utilization due to suboptimal configurations by users and the tendency to encounter abnormalities due to an unstable cloud environment. To overcome them, we introduce DLRover, an elastic training framework for DLRMs designed to increase resource utilization and handle the instability of a cloud environment. DLRover develops a resource-performance model by considering the unique characteristics of DLRMs and a three-stage heuristic strategy to automatically allocate and dynamically adjust resources for DLRM training jobs for higher resource utilization. Further, DLRover develops multiple mechanisms to ensure efficient and reliable execution of DLRM training jobs. Our extensive evaluation shows that DLRover reduces job completion times by 31%, increases the job completion rate by 6%, enhances CPU usage by 15%, and improves memory utilization by 20%, compared to state-of-the-art resource scheduling frameworks. DLRover has been widely deployed at AntGroup and processes thousands of DLRM training jobs on a daily basis. DLRover is open-sourced and has been adopted by 10+ companies.

Finding the future in digitally mediated ruin: #nostalgiacores and the algorithmic culture of digital platforms

The #nostalgiacores are a series of interrelated hashtags on Instagram and TikTok where users recirculate content from the digital and consumer cultures of the 1990s and 2000s – childhood play centres, dead malls, long-gone toys, and superseded game consoles and phones. In this article, we explore these digital cultures using a critical platform studies approach that involves a combination of network analysis and close textual analysis augmented with purpose-built machine vision tools. We scrape a collection of 359,150 images from Instagram that used one or more of 30 ‘-cores’ hashtags (such as #y2kcore, #webcore and #childhoodcore) that we chose following a period of immersive qualitative investigation of #nostalgiacore scenes on Instagram during 2021 and 2022. 10,000 Instagram images were then randomly selected and processed using a purpose-built unsupervised machine vision model that clusters images together based on their similarities. This research is part of a multi-year project where we develop hybrid digital methods for critically simulating and exploring the interplay between our image-making practices and the algorithmic systems that cluster and curate them. By combining computational approaches with critical platform and cultural studies approaches we speculatively explore both practices of curation and their interplay with the algorithmic classification and recommendation models of digital platforms. Our platform-oriented mode of textual analysis helps us to explore how our digital cultures are both symbolically and technically nostalgic. Instagram users in the #nostalgiacore scene recirculate images from the past as part of practices of critically reflecting on digital platforms and consumer cultures. At the same time those images are recuperated as archives used to train the algorithmic models that optimise attention on digital media platforms like Instagram.

Multi-stakeholder recommendation system through deep learning-based preference evaluation and aggregation model with multi-view information embedding

Learning the preferences of consumers, providers, and system stakeholders is a challenging problem in the Multi-Stakeholder Recommendation System (MSRS). Existing MSRS methods lack the ability to generate equitable recommendations and investigate implicit relationships between stakeholders and items. Hence, this study addresses this issue by proposing a multi-stakeholder preference learning-based recommendation model that exploits information from multiple views to evaluate stakeholders' preferences. The proposed model learns consumer preferences using users' ratings and reviews of an item, provider preferences with provider utility, and provider-item interaction. Furthermore, the proposed model learns the system-level preference of promoting long-tail items through the probabilistic evaluation of stakeholders' interest in popular and unpopular items. Finally, this study develops a multi-stakeholder, multi-view deep neural network model to aggregate stakeholders' preferences and deliver equitable recommendations. This work utilizes benchmark Movie Lens (ML) 25M, ML-100K, ML-1M, and TripAdvisor datasets to validate and compare the proposed model's performance with other baseline methods using standard evaluation metrics for each stakeholder. Examining the precision metrics, the proposed model attains the minimum enhancement of 7.91%, 18.24%, 10.72%, and 20.12% across the ML-25M, ML-100K, ML-1M, and TripAdvisor datasets. Further concerning the exposure, hit, and reach metrics, the model exhibits a substantial minimum improvement of 19.12%, 14.73%, 5.37%, and 28.46% over the ML-25M, ML-100K, ML-1M, and TripAdvisor datasets. Finally, the proposed model excels in promoting long-tail items and enhancing the cumulative utility gain of the stakeholders, surpassing the baseline methods.

Research and Application of a Dual Filtering Music Hybrid Recommendation Model Based on Cat Boost Algorithm and DCN

Research and Application of a Dual Filtering Music Hybrid Recommendation Model Based on Cat Boost Algorithm and DCN

Diabetes Prediction and Recommendation Model Using Machine Learning Techniques and MapReduce

Objectives: To deliver patient centric healthcare for diabetic patients using a fast and efficient diabetic prediction and recommendation model which will not only help in early diagnoses of disease but also recommend appropriate medicine for controlling it at stage 1. Methods: The Support Vector Machine Classifier is further enhanced with Particle Swarm Optimization (PSO) and used for the prediction of diabetes. Collaborative Filtering is used for drug recommendation, which produces a suitable list of medications that correspond to the diagnoses of diabetes patients. Improved Density-Based Spatial Clustering of Applications with Noise (I-DBSCAN) is proposed to cluster EHR data to get labels based on the symptoms of patients and map reduction is utilized to process the clustered data in parallel for quick recommendations. Findings: The accuracy of the SVM with the PSO model is 99.20%. The performance of I-DBSCAN is also compared with K-Means and regular DBSCAN using the Silhouette Score, Davies Bouldin Score, and the Calinski Harabasz Score. Also, I-DBSCAN was found to give a more accurate score. Novelty: The extensive volume of diabetes-related information stored in electronic health records (EHRs) through continuous monitoring devices poses a growing difficulty for healthcare professionals to effectively navigate and deliver patient-centered care. Machine Learning techniques like classification and recommendations can be utilized to facilitate early disease diagnosis and recommend appropriate medications. Keywords: Electronic health records (EHRs), Collaborative Filtering (CF), Recommendations, Improved Density Based Spatial Clustering of Applications with Noise (IDBSCAN), SVM classifier

Heterogeneous Meta-Path Graph Learning for Higher-Order Social Recommendation

Recommendation systems have become an indispensable part of daily life. Social recommendation systems, which utilize social relationships and past behaviors to infer users’ preferences, have gained popularity in recent years. Exploring the inherent characteristics implied by higher-order relationships offers a new approach to social recommendation. However, it is challenging due to sparse social networks, influence heterogeneity, and noisy feedback. In this article, we propose a Heterogeneous Meta-path Graph Learning model for Higher-order Social Recommendation (HEAL). Within HEAL, we introduce a heterogeneous graph in social recommendation and utilize a meta-path-guided random walk to generate higher-order relationships. By encoding higher-order structures and semantics along different meta-graphs, HEAL can mitigate the limitation of data sparsity. Moreover, HEAL exploits aspect-aware and semantic-aware attentions to adaptively propagate and aggregate useful features from different meta-neighbors and higher-order relations. These attention-based aggregation layers allow HEAL to suppress the heterogeneity of social influences. Furthermore, HEAL adopts contrastive learning as a supplemental task to the recommendation task by maximizing the consistency between the self-discriminating objectives. This auxiliary task enables the model to learn more differentiated representations, further reducing its sensitivity to noisy feedback. We evaluate the performance of HEAL through extensive experiments on public datasets. The results demonstrate that leveraging higher-order relations can enhance the quality of social recommendations by better capturing the complexity and diversity of users’ preferences and interactions.

Development of an Operating System Supporting Intelligent Predictions and Recommendations

This Study discusses the development of an intelligent operating system feature that supports smart prediction and recommendations using artificial intelligence (AI) capabilities within the Linux operating system. The study aims to integrate AI-driven features into Linux to enhance user productivity and efficiency by providing relevant application recommendations based on user behavior patterns. The implementation involves data collection of application usage, training machine learning models for application recommendations, and integrating these features into the Linux environment. The project utilizes Python for scripting, employing libraries such as psutil, pandas, scikit-learn, and joblib for data handling and machine learning tasks. The results demonstrate successful implementation of the AI-driven recommendation system, enhancing user interaction and productivity within the Linux operating system

Search-based Time-aware Graph-enhanced Recommendation with Sequential Behavior Data

Extending from sequential recommendation models, in this article, we present a novel framework named Search-based Time-Aware Recommendation (STARec), which first retrieves the historical behaviors of the given user through a search-based retriever and then captures the user’s evolving demands over time through a time-aware sequential network. We notice that the key insight of STARec is to use the feature and labels to augment the representations, and thus the effectiveness of STARec relies on the acquisition of rich browsing records of the target user and powerful representation of each browsed item and thus its performance could heavily drop regarding long-tail users and items. To this end, we extend STARec by constructing a graph upon the user–item interactions and leveraging the graph structure to enhance the representation learning. We call this extended version Search-based Time-Aware Graph-Enhanced Recommendation (STAGE). We conduct extensive experiments on three real-world datasets and STARec achieves consistent superiority. We further compare STAGE against STARec long-tail users and our results demonstrate that STAGE could outperform STARec at most cases. Results of online A/B tests show that STARec and STAGE achieve an average click-through rate improvement of around 6% and 1.5% in the two main item recommendation scenarios, respectively. 1

Artificial intelligence-based personalized clinical decision-making for patients with localized prostate cancer: surgery versus radiotherapy.

Surgery and radiotherapy are primary nonconservative treatments for prostate cancer (PCa). However, personalizing treatment options between these treatment modalities is challenging due to unclear criteria. We developed an artificial intelligence (AI)-based model that can identify patients with localized PCa who would benefit more from either radiotherapy or surgery, thereby providing personalized clinical decision-making. Data from consecutive patients with localized PCa who received radiotherapy or surgery with complete records of clinicopathological variables and follow-up results in 12 registries of the Surveillance, Epidemiology, and End Results database were analyzed. Patients from 7 registries were randomly assigned to training (TD) and internal validation datasets (IVD) at a 9:1 ratio. The remaining 5 registries constituted the external validation dataset (EVD). TD was divided into training-radiotherapy (TRD) and training-surgery (TSD) datasets, and IVD was divided into internal-radiotherapy (IRD) and internal-surgery (ISD) datasets. Six models for radiotherapy and surgery were trained using TRD and TSD to predict radiotherapy survival probability (RSP) and surgery survival probability (SSP), respectively. The models with the highest concordance index (C-index) on IRD and ISD were chosen to form the final treatment recommendation model (FTR). FTR recommendations were based on the higher value between RSP and SSP. Kaplan-Meier curves were generated for patients receiving recommended (consistent group) and nonrecommended treatments (inconsistent group), which were compared using the log-rank test. The study included 118 236 patients, categorized into TD (TRD: 44 621; TSD: 41 500), IVD (IRD: 4949; ISD: 4621), and EVD (22 545). Both radiotherapy and surgery models accurately predicted RSP and SSP (C-index: 0.735-0.787 and 0.769-0.797, respectively). The consistent group exhibited higher survival rates than the inconsistent group, particularly among patients not suitable for active surveillance (P < .001). FTR accurately identifies patients with localized PCa who would benefit more from either radiotherapy or surgery, offering clinicians an effective AI tool to make informed choices between these 2 treatments.

Efficient Exploration and Exploitation for Sequential Music Recommendation

Music streaming services heavily rely upon recommender systems to acquire, engage, and retain users. One notable component of these services are playlists, which can be dynamically generated in a sequential manner based on the user’s feedback during a listening session. Online learning to rank approaches have recently been shown effective at leveraging such feedback to learn users’ preferences in the space of song features. Nevertheless, these approaches can suffer from slow convergence as a result of their random exploration component and their session-agnostic exploitation component. To overcome these limitations, we propose a novel online learning to rank approach which efficiently explores the space of candidate recommendation models by restricting itself to the orthogonal complement of the subspace of previous underperforming exploration directions. Moreover, we propose a session-aware exploitation component which leverages the momentum of the current best model during updates. Our thorough evaluation using simulated listening sessions from two large Last.fm datasets demonstrates substantial improvements over state-of-the-art approaches in terms of early-stage performance, which results in an improved user experience during online learning. In addition, we demonstrate that long-term convergence can be further enhanced by adaptively relaxing exploration constraints along the way.

Constrained Off-policy Learning over Heterogeneous Information for Fairness-aware Recommendation

Fairness-aware recommendation eliminates discrimination issues to build trustworthy recommendation systems. Existing fairness-aware approaches ignore accounting for rich user and item attributes and thus cannot capture the impact of attributes on affecting recommendation fairness. These real-world attributes severely cause unfair recommendations by favoring items with popular attributes, leading to item exposure unfairness in recommendations. Moreover, existing approaches mostly mitigate unfairness for static recommendation models, e.g., collaborative filtering. Static models can not handle dynamic user interactions with the system that reflect users’ preferences shift through time. Thus, static models are limited in their ability to adapt to user behavior shifts to gain long-run user satisfaction. As user and item attributes are largely involved in modern recommenders and user interactions are naturally dynamic, it is essential to develop a novel method that eliminates unfairness caused by attributes meanwhile embrace the dynamic modeling of user behavior shifts. In this article, we propose Constrained Off-policy Learning over Heterogeneous Information for Fairness-aware Recommendation (Fair-HINpolicy) , which uses recent advances in context-aware off-policy learning to produce fairness-aware recommendations with rich attributes from a Heterogeneous Information Network. In particular, we formulate the off-policy learning as a Constrained Markov Decision Process (CMDP) by dynamically constraining the fairness of item exposure at each iteration. We also design an attentive action sampling to reduce the search space for off-policy learning. Our solution adaptively receives HIN-augmented corrections for counterfactual risk minimization, and ultimately yields an effective policy that maximizes long-term user satisfaction. We extensively evaluate our method through simulations on large-scale real-world datasets, obtaining favorable results compared with state-of-the-art methods.

Disentangled Cascaded Graph Convolution Networks for Multi-Behavior Recommendation

Multi-behavioral recommender systems have emerged as a solution to address data sparsity and cold-start issues by incorporating auxiliary behaviors alongside target behaviors. However, existing models struggle to accurately capture varying user preferences across different behaviors and fail to account for diverse item preferences within behaviors. Various user preference factors (such as price or quality) entangled in the behavior may lead to sub-optimization problems. Furthermore, these models overlook the personalized nature of user behavioral preferences by employing uniform transformation networks for all users and items. To tackle these challenges, we propose the Disentangled Cascaded Graph Convolutional Network (Disen-CGCN), a novel multi-behavior recommendation model. Disen-CGCN employs disentangled representation techniques to effectively separate factors within user and item representations, ensuring their independence. In addition, it incorporates a multi-behavioral meta-network, enabling personalized feature transformation across user and item behaviors. Furthermore, an attention mechanism captures user preferences for different item factors within each behavior. By leveraging attention weights, we aggregate user and item embeddings separately for each behavior, computing preference scores that predict overall user preferences for items. Our evaluation of benchmark datasets demonstrates the superiority of Disen-CGCN over state-of-the-art models, showcasing an average performance improvement of 7.07% and 9.00% on respective datasets. These results highlight Disen-CGCN’s ability to effectively leverage multi-behavioral data, leading to more accurate recommendations.

A Hybrid Recommendation Model for Tourist Using Evolutionary Algorithm Combined with Local Search Algorithm for Trip Planning

A Hybrid Recommendation Model for Tourist Using Evolutionary Algorithm Combined with Local Search Algorithm for Trip Planning

Decentralized Federated Recommendation with Privacy-aware Structured Client-level Graph

Recommendation models are deployed in a variety of commercial applications to provide personalized services for users. However, most of them rely on the users’ original rating records that are often collected by a centralized server for model training, which may cause privacy issues. Recently, some centralized federated recommendation models are proposed for the protection of users’ privacy, which however requires a server for coordination in the whole process of model training. As a response, we propose a novel privacy-aware decentralized federated recommendation (DFedRec) model, which is lossless compared with the traditional model in recommendation performance and is thus more accurate than other models in this line. Specifically, we design a privacy-aware structured client-level graph for the sharing of the model parameters in the process of model training, which is a one-stone-two-bird strategy, i.e., it protects users’ privacy via some randomly sampled fake entries and reduces the communication cost by sharing the model parameters only with the related neighboring users. With the help of the privacy-aware structured client-level graph, we propose two novel collaborative training mechanisms in the setting without a server, including a batch algorithm DFedRec(b) and a stochastic one DFedRec(s), where the former requires the anonymity mechanism while the latter does not. They are both equivalent to probabilistic matrix factorization trained in a centralized server and are thus lossless. We then provide formal analysis of privacy guarantee of our methods and conduct extensive empirical studies on three public datasets with explicit feedback, which show the effectiveness of our DFedRec, i.e., it is privacy aware, communication efficient, and lossless.

Inclusion of people with disability (PWD) in the tourism workforce: a critical discourse analysis of policies and plans

Purpose This paper aims to critically examine the discourse on the disability workforce in Australian tourism policy and planning documents. Design/methodology/approach This research draws on a database of 490 Australian tourism policy and planning documents, across national, state, regional and local levels of governments, destination management organisations and peak industry bodies, to systematically examine the issues pertaining to disability workforce over the ten-year policy cycle (2013–2023), through critical discourse analysis. Findings Analysis revealed 20 of the existing 490 documents had mentions of the inclusion of people with disability (PWD) in the tourism workforce. Through critical discourse analysis, the nuances of political attitudes are evident. Tourism policies tend to exert a cautious approach towards the inclusion of PWD employment. PWD do not have a strong voice when it comes to the formation of tourism policy, especially not as members of the tourism workforce. Practical implications This research underscores an emerging call for the government’s commitment to considering the underrepresented social group of PWD in the tourism workforce. By shedding light on critical discourse analysis of the disability workforce, tourism scholars and practitioners understand the current challenges and explore opportunities to implement tangible actions for an inclusive work environment for all. Originality/value The research is novel and innovative as it critically examines the discourse of the disability workforce in the tourism sector through an institutional theory perspective. The limited emphasis of government policies on the disability workforce may result in missed opportunities to encourage meaningful employment engagement of PWD in tourism organisations. Consequently, this research contributes a tripartite model of policy recommendations that explicitly articulates the mechanisms for change that are tailored to tourism workforce and advocates for the active voices of PWD community in the policy formation process.

Multimodal Pre-training for Sequential Recommendation via Contrastive Learning

Sequential recommendation systems often suffer from data sparsity, leading to suboptimal performance. While multimodal content, such as images and text, has been utilized to mitigate this issue, its integration within sequential recommendation frameworks remains challenging. Current multimodal sequential recommendation models are often unable to effectively explore and capture correlations among behavior sequences of users and items across different modalities, either neglecting correlations among sequence representations or inadequately capturing associations between multimodal data and sequence data in their representations. To address this problem, we explore multimodal pre-training in the context of sequential recommendation, with the aim of enhancing fusion and utilization of multimodal information. We propose a novel M ultimodal P re-training for S equential R ecommendation (MP4SR) framework, which utilizes contrastive losses to capture the correlation among different modality sequences of users, as well as the correlation among different modality sequences of users and items. MP4SR consists of three key components: 1) multimodal feature extraction, 2) a backbone network, Multimodal Mixup Sequence Encoder (M 2 SE), and 3) pre-training tasks. After utilizing pre-trained encoders to generate initial multimodal features of items, M 2 SE adopts a complementary sequence mixup strategy to fuse different modality sequences, and leverages contrastive learning to capture modality interactions at the sequence-to-sequence and sequence-to-item levels. Extensive experiments on four real-world datasets demonstrate that MP4SR outperforms state-of-the-art approaches in both normal and cold-start settings. We further highlight the efficacy of incorporating multimodal pre-training in sequential recommendation representation learning, serving as an effective regularizer and optimizing the parameter space for the recommendation task.

Boosting Healthiness Exposure in Category-Constrained Meal Recommendation Using Nutritional Standards

Food computing, a newly emerging topic, is closely linked to human life through computational methodologies. Meal recommendation, a food-related study about human health, aims to provide users a meal with courses constrained from specific categories (e.g., appetizers, main dishes) that can be enjoyed as a service. Historical interaction data, important user information, is often used by existing models to learn user preferences. However, if a user’s preferences favor less healthy meals, the model will follow that preference and make similar recommendations, potentially negatively impacting the user’s long-term health. This emphasizes the necessity for health-oriented and responsible meal recommendation systems. In this article, we propose a healthiness-aware and category-wise meal recommendation model called CateRec, which boosts healthiness exposure by using nutritional standards as knowledge to guide the model training. Two fundamental questions are raised and answered: (1) How can the healthiness of meals be evaluated? Two well-known nutritional standards from the World Health Organization and the United Kingdom Food Standards Agency are used to calculate the healthiness score of the meal. (2) How can the model training be guided in a health-oriented manner? We construct category-wise personalization partial rankings and category-wise healthiness partial rankings, and theoretically analyze that they meet the necessary properties and assumptions required to be trained by the maximum posterior estimator under Bayesian probability. The data analysis confirms the existence of user preferences leaning towards less healthy meals in two public datasets. A comprehensive experiment demonstrates that our CateRec effectively boosts healthiness exposure in terms of mean healthiness score and ranking exposure while being comparable to the state-of-the-art model in terms of recommendation accuracy.

Knowledge Graph Enhanced Contextualized Attention-Based Network for Responsible User-Specific Recommendation

With ever-increasing dataset size and data storage capacity, there is a strong need to build systems that can effectively utilize these vast datasets to extract valuable information. Large datasets often exhibit sparsity and pose cold start problems, necessitating the development of responsible recommender systems. Knowledge graphs have utility in responsibly representing information related to recommendation scenarios. However, many studies overlook explicitly encoding contextual information, which is crucial for reducing the bias of multi-layer propagation. Additionally, existing methods stack multiple layers to encode high-order neighbor information while disregarding the relational information between items and entities. This oversight hampers their ability to capture the collaborative signal latent in user-item interactions. This is particularly important in health informatics, where knowledge graphs consist of various entities connected to items through different relations. Ignoring the relational information renders them insufficient for modeling user preferences. This work presents an end-to-end recommendation framework named KGCAN (Knowledge Graph Enhanced Contextualized Attention-Based Network), which explicitly encodes both relational and contextual information of entities to preserve the original entity information. Furthermore, a user-specific attention mechanism is employed to capture personalized recommendations. The proposed model is validated on three benchmark datasets through extensive experiments. The experimental results demonstrate that KGCAN outperforms existing knowledge graph based recommendation models. Additionally, a case study from the healthcare domain is discussed, highlighting the importance of attention mechanisms and high-order connectivity in the responsible recommendation system for health informatics.

Data Collaborative Contrastive Recommendation model with self-adaptive noise

The recommender system recommends items to the users based on their preferences of implicit feedback. However, implicit feedback often contains noise that deviates from the user’s true preferences, thereby influencing the accuracy of the recommendations. The most effective denoising method is Self-Guided Denoising Learning (SGDL), providing a general denoising scheme that can be applied to various recommendation models. However, it is typically hard to capture user preferences efficiently and enhance the diversity of recommendations to mitigate the ‘filter bubble’ phenomenon. To address these challenges, we propose a novel Data Collaborative Contrastive Recommendation model with self-adaptive noise (DCCR). Specifically, we design the informative item extraction module to mine informative items from the original interactions and improve the accuracy and diversity of recommendations by collaborative training of the informative and the original dataset to learn diverse user embeddings and adapt to noise. Extensive experiments on three public datasets demonstrate the superiority of our DCCR over state-of-the-art methods, balancing the diversity of recommendation lists while optimizing recommendation accuracy.

A GAI-based multi-scale convolution and attention mechanism model for music emotion recognition and recommendation from physiological data

Due to the subjectivity of emotions and the limited number of emotion categories, existing deep learning models require assistance to achieve objective, accurate, and flexible personalized music emotion recommendations. This paper introduces a deep learning approach that combines Generative Artificial Intelligence (GAI) and explicitly leverages physiological indicators to enhance the model's intelligence, versatility, and automation. Physiological indicators such as Heart Rate Variability (HRV) and Galvanic Skin Response (GSR) can be measured using sensors placed on the body's surface, providing more precise information about human emotional changes. This research employs a three-dimensional emotion model, including the tension-arousal axis, energy-arousal axis, and valence axis, to explain the correlation and accuracy between music data and emotions. Based on this, a music emotion classifier is designed, incorporating GAI algorithms to recommend music by matching users' physiological and emotional types with the emotional features of music. The classifier uses Mel-Frequency Cepstral Coefficients (MFCC) to transform audio into Mel-spectrogram as input features. The music emotion selection module adopts a GAI framework of Variational Autoencoder (VAE) and integrates multi-scale parallel convolution and attention mechanism modules. Experimental results demonstrate that this approach is competitive compared to existing deep learning architectures on PMEmo, RAVDESS, and Soundtrack datasets. Furthermore, due to GAI's efficient classification capability, this model is suitable for resource-constrained mobile devices and other smart devices. The results of this study can be applied to emotion-based music recommendation systems, contributing to emotional interventions and improving the performance of exercise and music therapy.