Search
Search Funnelback University
- Refined by:
- Date: 2013
1 -
16 of
16
search results for news |u:mi.eng.cam.ac.uk
Fully-matching results
-
Non-Rigid Photometric Stereo with Colored Lights
mi.eng.cam.ac.uk/research/projects/VideoNormals/23 Apr 2013: Old New [41 Mb mpeg4 avi]. ... Short Small [5.7 Mb mpg1]. Results (ICCV07). New Results. Data. Contacts. -
Engineering Tripos Part IIB FOURTH YEAR Module 4F10: STATISTICAL ...
mi.eng.cam.ac.uk/~mjfg/local/4F10/examples1.2up.pdf8 Nov 2013: xn are used to train the mixture model. Usingthe standard form of EM with mixture models show that the maximum likelihoodestimate for the “new” parameters, λ̂ji, is given by. -
University of CambridgeEngineering Part IB Paper 8 Information…
mi.eng.cam.ac.uk/~cipolla/lectures/PartIB/old/2013-IB-handout5.pdf14 May 2013: new word belongs to. ... Second, there is an initialprocess of learning which enables the technique tothen categorize new, unseen images. -
paper.dvi
mi.eng.cam.ac.uk/~mjfg/flego_icassp12.pdf13 Jun 2013: 2.1. Maximum Likelihood trainingTo estimate the new VAT canonical model parametersM̂ thefollowing auxiliary function can be used. ... 3 and 4, is maximised, anda back-off procedure on the new estimates is generally ap-plied [2]. -
IB-search.dvi
mi.eng.cam.ac.uk/~cipolla/lectures/PartIB/old/2013-IB-handout4.pdf14 May 2013: return new Node(value, nothing, nothing);. if(value == current.value). return current;. if(value < current.value). ... returns KDNode. {. if(data.Count == 1). return new KDNode(nothing, -1, 0, data[0],. -
Infinite Support Vector Machines in Speech Recognition Jingzhou Yang, …
mi.eng.cam.ac.uk/~mjfg/yang_is2013.pdf13 Jun 2013: α. i α, where m is a new expert. (9). According to equation (9), the probability of the indicator indi-cating an existing component is proportional to the number ofdata associated ... with that component, and the probability of as-signing to a new -
paper.dvi
mi.eng.cam.ac.uk/~mjfg/wang_icassp13.pdf13 Jun 2013: Thus the ability to adapt ASRsystems to new operating conditions, unseen in the trainingdata, is important. -
paper.dvi
mi.eng.cam.ac.uk/~mjfg/wang_is2012.pdf13 Jun 2013: A new adaptive train-ing algorithm, reverberant adaptive training (RAT), is proposed,where the RVTSJ adaptation is used in both testing and trainingstage. -
KERNELIZED LOG LINEAR MODELS FOR CONTINUOUS SPEECH RECOGNITION…
mi.eng.cam.ac.uk/~mjfg/Kernel/sxz20_icassp13.pdf27 Mar 2013: New York, NY, USA: Cambridge University Press, 2004. [14] Y. Kubo, S. ... New York:Springer-Verlag, 1995. [18] J. Weston, B. Schölkopf, and O. Bousquet, “Joint kernel maps,” inIWANN, 2005, pp. -
A HIGH-PERFORMANCE CANTONESE KEYWORD SEARCH SYSTEM
mi.eng.cam.ac.uk/~mjfg/ICASSP13_ibm2.pdf13 Jun 2013: Nhyp(T ), then sum-to-one normalization computes new scores. ŝ(i,T ) = s(i,T )/Nhyp(T )X. ... Tranter, “Progress in the CU-HTKBroadcast News transcription system,” IEEE Transactions onAudio, Speech, and Language Processing, vol. -
INFERENCE ALGORITHMS FOR GENERATIVE SCORE-SPACES A. Ragni and M. ...
mi.eng.cam.ac.uk/~mjfg/Kernel/ragni_icassp12.pdf27 Mar 2013: αTφ(Oτ :t, ωm) = α. (ωm) log (p(Oτ :t|ωm)) (13). Equations (12) and (13) suggest that each time new segmentationθτ :t is considered the likelihood has to be -
A CONFIDENCE-BASED APPROACH FOR IMPROVING KEYWORD HYPOTHESIS SCORES…
mi.eng.cam.ac.uk/~mjfg/seigel_ICASSP2013.pdf13 Jun 2013: In thisprocess, sub-words which form part of different word-level hypothe-ses can be grouped together as part of new key term hypotheses thatare not present in the lattice. ... James, “A system for unrestricted topic retrieval fromradio news broadcasts, -
SYSTEM COMBINATION AND SCORE NORMALIZATION FOR SPOKEN TERM DETECTION
mi.eng.cam.ac.uk/~mjfg/ICASSP13_ibm1.pdf13 Jun 2013: We apply these methods to Cantonese,a language that presents some new issues in terms of reduced re-sources and shorter query lengths. ... This approach is inspired from [8]. Sum-to-one Normalization (STO): sum-to-one normalizationcomputes new scores as. -
draft21.dvi
mi.eng.cam.ac.uk/~mjfg/ASRU13.pdf7 Nov 2013: benefit of simply requiring the output layer to be trained for a new. ... Prior to recognition, Tandem GMM-HMMs must be trained based on. the new Tandem features as described in Section 3.2. -
Department of Engineering 1 Generative Kernels and Score-Spaces…
mi.eng.cam.ac.uk/~mjfg/Kernel/rcv25_2013_y2.pdf9 Sep 2013: Because of the nature of the Dirichlet process, the posteriordistribution of the assignment of one observation given all other ones is split betweenthe existing components and the new ones:. ... P(zn = k. zn,η) {Nk, if k is an existing expert;η, if k -
Department of Engineering 1 E�cient decodingwith continuous rational…
mi.eng.cam.ac.uk/~mjfg/Kernel/van_dalen-2012-tr-efficient_score-spaces.pdf27 Mar 2013: e semiring operations dened on these new weights can be derived in various ways.e simplest for the purpose of this paper is to describe the derivatives of the sum orproduct
Search history
Recently clicked results
Recently clicked results
Your click history is empty.
Recent searches
Recent searches
Your search history is empty.