Latent factor space and leverage latent factors to

Info icon This preview shows pages 10–12. Sign up to view the full content.

latent factor space and leverage latent factors to explain the generation of QoS values. In this paper, we jointly model the network location-aware neighbor and regularized MF. We also assume q ¼ x to avoid the over¯tting problem, and the intuition is to reduce the number of parameters. ^ r ui ¼ " þ b u þ b i þ p T u q i þ 1 ffiffiffiffiffiffiffiffiffiffi R ð u Þ p x T i X j 2 R ð u Þ y j ¼ " þ b u þ b i þ q T i p u þ 1 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R ð AS ð i Þ ; u Þ p X j 2 R ð AS ð i Þ ; u Þ y j 0 @ 1 A : ð 9 Þ Similar to the MF model, we also employ regularization terms to avoid the over- ¯tting problem. The proposed service location-aware MF model (SLMF) is given as L ¼ 1 2 X ð u ; i Þ2 R r ui ³ " ³ b u ³ b i ³ q T i p u þ 1 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R ð AS ð i Þ ; u Þ p X j 2 R ð AS ð i Þ ; u Þ y j 0 @ 1 A 0 @ 1 A 2 þ ! 2 b 2 u þ b 2 i þ jj q i jj 2 þ jj p u jj 2 þ X j 2 R ð AS ð i Þ ; u Þ jj y j jj 2 0 @ 1 A ; ð 10 Þ a . 620 Y. Yin et al. Int. J. Soft. Eng. Knowl. Eng. 2016.26:611-632. Downloaded from by WEIZMANN INSTITUTE OF SCIENCE on 07/01/16. For personal use only.
Image of page 10

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

where jj µ jj 2 is the Frobenius norm and ! is a small constant. The impact of the setting of ! to our model will be studied in the experimental section. We name this model service location-aware MF model (SLMF). Notice that, in our model, the term location refers to the network location, instead of the geographic location. Mean- while, we construct a service-network aware regularization term P j 2 R ð AS ð i Þ ; u Þ jj y j jj 2 . We also use the gradient descent (see Eq. ( 5 )) to achieve the local optima of all parameters. In Eq. ( 10 ), we give our ¯rst model, i.e. the MF model extended from service location-aware neighbor selection. Similarly, according to the explanation in Sec. 1 , we also have the observation that users in the same network are likely to receive similar QoS values. Based on the network location-aware neighbor selection in Sec. 4.3 , we give the second proposed model below, i.e. the MF model extended from user location-aware neighbor selection. ^ r ui ¼ " þ b u þ b i þ p T u q i þ 1 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R ð AS ð u Þ ; i Þ p X v 2 R ð AS ð u Þ ; i Þ z v 0 @ 1 A ; ð 11 Þ where z v represents the latent factors of the users that are in the same network with the target user u . The objective function of our model is given as L ¼ 1 2 X ð u ; i Þ2 R r ui ³ " ³ b u ³ b i ³ p T u q i þ 1 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi R ð AS ð u Þ ; i Þ p X v 2 R ð AS ð u Þ ; i Þ z v 0 @ 1 A 0 @ 1 A 2 þ ! 2 b 2 u þ b 2 i þ jj q i jj 2 þ jj p u jj 2 þ X v 2 R ð AS ð u Þ ; i Þ jj z v jj 2 0 @ 1 A : ð 12 Þ We name this model user location-aware MF model (ULMF). Meanwhile, in the ULMF model, we construct a user network-aware regularization term P v 2 R ð AS ð u Þ ; i Þ jj z v jj 2 . Also, the term location refers to the network location. We can also use the gradient descent
Image of page 11
Image of page 12
This is the end of the preview. Sign up to access the rest of the document.
  • Winter '15
  • MAhmoudali
  • Analysis of algorithms, Computational complexity theory, Service system, Expectation-maximization algorithm, Weizmann Institute of Science

{[ snackBarMessage ]}

What students are saying

  • Left Quote Icon

    As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

    Student Picture

    Kiran Temple University Fox School of Business ‘17, Course Hero Intern

  • Left Quote Icon

    I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

    Student Picture

    Dana University of Pennsylvania ‘17, Course Hero Intern

  • Left Quote Icon

    The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

    Student Picture

    Jill Tulane University ‘16, Course Hero Intern