Advances, Systems and Applications
From: QoS prediction in intelligent edge computing based on feature learning
Notation | Description |
---|---|
\(es\in {ES}\) | the edge sever |
\(nm\in {NM}\) | the network mode |
\({u_i}\in {U}\) | a user, \(i\in {\{1,\cdots ,m\}}\) |
\({s_j}\in {S}\) | a service, \(j\in {\{1,\cdots ,n\}}\) |
m | the number of users |
n | the number of services |
K | the number of user device types |
s | the number of field(explicit feature) |
d | interaction layer input vector dimension |
\(em\in {EM}\) | explicit embedding vector |
\(Q\in {{R}^{m\times n}}\) | QoS matrix |
\(U_{IM}\in {{R}^{m\times d}}\) | user implicit feature matrix |
\(S_{IM}\in {{R}^{n\times d}}\) | sevice implicit feature matrix |
\({\lambda }_{reg}\) | regularization term |
\({U}_{I{{M}_{i}}}\) | implicit feature of user i |
\({S}_{I{{M}_{j}}}\) | implicit feature of service j |
\({e_f}\in {e}\) | feature used to interact |
\(\alpha _{f,k}^{\left( h \right) }\) | attention weights between feature f and k |
\(\tilde{e}_{f}^{(h)}\) | a new combinatorial feature under head h |
H | the number of head |