IVAS: Facilitating Safe and Comfortable Driving with Intelligent Vehicle Audio Systems

Wireless recommendations for Internet of vehicles: Recent advances, challenges, and opportunities

“…In user-based models [23] , collaborative filtering algorithm aims at filtering out items that a user might like on the basis of reactions by similar users. In content/item-based User-based method [23] Item-based method [24--26] -Similarities evaluation [27--29] -Ability to recommend similar items -Ability to learn from similar users -Personalized recommendation -Sparsity -Cold-start -Lack of diversity [30] -Matrix factorization [31,32] SVD++ [33] , SVD [34] , and FM [35] -Feedback and side information [36,37] -Better representations with dense embeddings and better rating estimation with the inner product -Poor inner-product-based predication [38,39] -Ambiguous representations [40] Deep learningbased technique -Neural CF [40--43] -Deep neural network [44] -GNN and graph embedding [45,46] - -Contextual MAB [47--49] -Markov Decision Process [50] -Upper Confidence Bound [51--53] -Thompson Sampling [54] - models [24--26] , collaborative filtering algorithm tries to predict a user's interest on an item based on the user's interests on similar items/contents. One key question is how to evaluate the similarities between users or items.…”

“…In Ref. [44], an intelligent vehicle audio system in the IoV was proposed to make driving strategy recommendations based on deep learning techniques. More recently, graph embedding [45] and graph neural networks [46] have also attention in the design of recommendation systems due to that they are more suitable to capture discrete or sequential inputs.…”