Search
Search Funnelback University
- Refined by:
- Date: 2018
1 -
10 of
90
search results for b&b |u:mi.eng.cam.ac.uk
Fully-matching results
-
hierParsing.dvi
mi.eng.cam.ac.uk/~sjy/papers/heyo03a.pdf20 Feb 2018: b."B-}b@_ ¤& }M¡¤&¢}ª|&}b|&bjD«« },£ ¤&9}¡¥}b. @}b}c }Xj}bXÎjU9}¡»-}b}b0¢@}X XB} j¡}M@_Í»j%¢|& j }M_lË& ... aB}Ã&¤|&}M j}Mj|&}Xö-jís"@}X j&}jÎ?|}b|&}_}M¢¤&@Ëj@}b }769}M[@}b$ -
Reward-Balancing for Statistical Spoken Dialogue Systems…
mi.eng.cam.ac.uk/~sjy/papers/ubcy17.pdf20 Feb 2018: In dialogue management, the following kernelhas been successfully used:. k((b,a),(b′,a′)) = δ(a,a′) klin(b,b′). ... k((b,a,w),(b′,a′,w′)). = δ(a,a′) (klin(b,b. ′) klin(w,w′)). (9). 1A similar type of kernel extension has been -
Incremental on-line adaptation of POMDP-based dialogue managers…
mi.eng.cam.ac.uk/~sjy/papers/gktb14.pdf20 Feb 2018: k(, )is factored into separate kernels over the belief state and ac-tion space kB(b,b′)kA(a,a′). ... kB(b,b′) =. h. 〈bh,b′h〉, (6). where bh is the probability distribution encoded in the hth hid-den node. -
DISTRIBUTED DIALOGUE POLICIES FOR MULTI-DOMAIN STATISTICAL…
mi.eng.cam.ac.uk/~sjy/papers/gkty15a.pdf20 Feb 2018: Based on this input, at every dialogue turn, a distributionof possible dialogue states called the belief state, an element of be-lief space b B, is estimated. ... kB(b,b′) =. h. 〈bh,b′h〉 (6). where bh is the probability distribution encoded in -
DISTRIBUTED DIALOGUE POLICIES FOR MULTI-DOMAIN STATISTICAL…
mi.eng.cam.ac.uk/~sjy/papers/gkty15.pdf20 Feb 2018: Based on this input, at every dialogue turn, a distributionof possible dialogue states called the belief state, an element of be-lief space b B, is estimated. ... kB(b,b′) =. h. 〈bh,b′h〉 (6). where bh is the probability distribution encoded in -
POLICY COMMITTEE FOR ADAPTATION IN MULTI-DOMAIN SPOKEN…
mi.eng.cam.ac.uk/~sjy/papers/gmsv15.pdf20 Feb 2018: Based on this input, at every di-alogue turn, a distribution of possible dialogue states calledthe belief state, an element of belief space b B, is esti-mated. ... kB(b,b′) =. h. 〈bh,b′h〉 (8). where bh is the probability distribution encoded in -
paper.dvi
mi.eng.cam.ac.uk/~sjy/papers/yswy07.pdf20 Feb 2018: u, sd, am)| {z }dialog model. P (p′|p)b(p, sd)| {z }belief refinement. ... m a. b. b. Summary Space. Fig. 3. Overview of Prototype HIS Dialog Manager. -
POLICY OPTIMISATION OF POMDP-BASED DIALOGUE SYSTEMS WITHOUT…
mi.eng.cam.ac.uk/~sjy/papers/ghtt12.pdf20 Feb 2018: It takes values b B, where Bis a continuous space of dimensionality |S|, namely [0, 1]|S|. ... kB(b,b′) =. k. 〈bk,b′k〉, (15). where bk is the probability distribution encoded in the kth hiddennode. -
Inverse Reinforcement Learning for Micro-Turn Management Dongho Kim,…
mi.eng.cam.ac.uk/~sjy/papers/kgbt14.pdf20 Feb 2018: P(s′|s,a) =P(m′|a)P(l′|l,m,m′,u,u′)P(b′|b,m,m′,u,u′)P(c′|c,m,m′,u,u′) ... 0.8. 1.0. Polic. y speaksilentbackchannel. (b) Overlap. 0.0 0.2 0.4 0.6 0.8 1.0Time (sec). -
THE USE OF DISCRIMINATIVE BELIEF TRACKINGIN POMDP-BASED DIALOGUE…
mi.eng.cam.ac.uk/~sjy/papers/khgt14.pdf20 Feb 2018: b′ = softmax([h g]{B}). where B is a parameter of the RNN. ... 5. 0. 5. 10. Ave. rage. rew. ard. RNNBNBN (constrained). (b) 30% error rate.
Search history
Recently clicked results
Recently clicked results
Your click history is empty.
Recent searches
Recent searches
Your search history is empty.