## William W. Cohen, Andrew Moore

*Proceedings of the 23rd International Conference on Machine Learning*

ICML, 2006.

@proceedings{ICML-2006, address = "Pittsburgh, Pennsylvania, USA", editor = "William W. Cohen and Andrew Moore", isbn = "1-59593-383-2", publisher = "{ACM}", series = "{ACM International Conference Proceeding Series}", title = "{Proceedings of the 23rd International Conference on Machine Learning}", volume = 148, year = 2006, }

### Contents (140 items)

- ICML-2006-AbbeelQN #learning #modelling #using
- Using inaccurate models in reinforcement learning (PA, MQ, AYN), pp. 1–8.
- ICML-2006-AgarwalHKS #algorithm
- Algorithms for portfolio management based on the Newton method (AA, EH, SK, RES), pp. 9–16.
- ICML-2006-AgarwalBB #graph #higher-order #learning
- Higher order learning with graphs (SA, KB, SB), pp. 17–24.
- ICML-2006-Agarwal #graph #ranking
- Ranking on graph data (SA0), pp. 25–32.
- ICML-2006-ArchambeauDV #probability #robust
- Robust probabilistic projections (CA, ND, MV), pp. 33–40.
- ICML-2006-ArgyriouHMP #algorithm #kernel
- A DC-programming algorithm for kernel selection (AA, RH, CAM, MP), pp. 41–48.
- ICML-2006-AsgharbeygiSL #difference #learning #relational
- Relational temporal difference learning (NA, DJS, PL), pp. 49–56.
- ICML-2006-AzranG #approach #clustering #data-driven
- A new approach to data driven clustering (AA, ZG), pp. 57–64.
- ICML-2006-BalcanBL #learning
- Agnostic active learning (MFB, AB, JL), pp. 65–72.
- ICML-2006-BalcanB #formal method #learning #on the #similarity
- On a theory of learning with similarity functions (MFB, AB), pp. 73–80.
- ICML-2006-Banerjee #bound #on the
- On Bayesian bounds (AB), pp. 81–88.
- ICML-2006-BanerjeeGdN #modelling #optimisation #visual notation
- Convex optimization techniques for fitting sparse Gaussian graphical models (OB, LEG, Ad, GN), pp. 89–96.
- ICML-2006-BeygelzimerKL #nearest neighbour
- Cover trees for nearest neighbor (AB, SK, JL), pp. 97–104.
- ICML-2006-BezakovaKS #graph #using
- Graph model selection using maximum likelihood (IB, AK, RS), pp. 105–112.
- ICML-2006-BleiL #modelling #topic
- Dynamic topic models (DMB, JDL), pp. 113–120.
- ICML-2006-BonillaWACTO #predict
- Predictive search distributions (EVB, CKIW, FVA, JC, JT, MFPO), pp. 121–128.
- ICML-2006-BowlingMJNW #learning #policy #predict #using
- Learning predictive state representations using non-blind policies (MHB, PM, MJ, JN, DFW), pp. 129–136.
- ICML-2006-BrefeldGSW #performance
- Efficient co-regularised least squares regression (UB, TG, TS, SW), pp. 137–144.
- ICML-2006-BrefeldS #learning
- Semi-supervised learning for structured output variables (UB, TS), pp. 145–152.
- ICML-2006-Carreira-Perpinan #clustering #parametricity #performance
- Fast nonparametric clustering with Gaussian blurring mean-shift (MÁCP), pp. 153–160.
- ICML-2006-CaruanaN #algorithm #comparison #empirical #learning
- An empirical comparison of supervised learning algorithms (RC, ANM), pp. 161–168.
- ICML-2006-CaytonD #robust
- Robust Euclidean embedding (LC, SD), pp. 169–176.
- ICML-2006-Cesa-BianchiGZ #classification
- Hierarchical classification: combining Bayes with SVM (NCB, CG, LZ), pp. 177–184.
- ICML-2006-ChapelleCZ #continuation
- A continuation method for semi-supervised SVMs (OC, MC, AZ), pp. 185–192.
- ICML-2006-CheungK #framework #learning #multi
- A regularization framework for multiple-instance learning (PMC, JTK), pp. 193–200.
- ICML-2006-CollobertSWB #scalability
- Trading convexity for scalability (RC, FHS, JW, LB), pp. 201–208.
- ICML-2006-ConitzerG #algorithm #learning #online #problem
- Learning algorithms for online principal-agent problems (and selling goods online) (VC, NG), pp. 209–216.
- ICML-2006-SilvaBBE #detection #using
- Dealing with non-stationary environments using context detection (BCdS, EWB, ALCB, PME), pp. 217–224.
- ICML-2006-DaiYTK #adaptation #classification #nondeterminism
- Locally adaptive classification piloted by uncertainty (JD, SY, XT, JTK), pp. 225–232.
- ICML-2006-DavisG
- The relationship between Precision-Recall and ROC curves (JD, MG), pp. 233–240.
- ICML-2006-TorreK #analysis #clustering
- Discriminative cluster analysis (FDlT, TK), pp. 241–248.
- ICML-2006-DeCoste #collaboration #matrix #predict #using
- Collaborative prediction using ensembles of Maximum Margin Matrix Factorizations (DD), pp. 249–256.
- ICML-2006-DegrisSW #learning #markov #problem #process
- Learning the structure of Factored Markov Decision Processes in reinforcement learning problems (TD, OS, PHW), pp. 257–264.
- ICML-2006-DenisMR #classification #learning #naive bayes #performance
- Efficient learning of Naive Bayes classifiers under class-conditional classification noise (FD, CNM, LR), pp. 265–272.
- ICML-2006-desJardinsEW #learning #set
- Learning user preferences for sets of objects (Md, EE, KW), pp. 273–280.
- ICML-2006-DingZHZ #analysis #component #invariant #named #robust
- R1-PCA: rotational invariant L1-norm principal component analysis for robust subspace factorization (CHQD, DZ, XH, HZ), pp. 281–288.
- ICML-2006-Elkan #approximate #clustering #documentation #multi
- Clustering documents with an exponential-family approximation of the Dirichlet compound multinomial distribution (CE), pp. 289–296.
- ICML-2006-EngelhardtJB #predict #visual notation
- A graphical model for predicting protein molecular function (BEE, MIJ, SEB), pp. 297–304.
- ICML-2006-EpshteynD #learning
- Qualitative reinforcement learning (AE, GD), pp. 305–312.
- ICML-2006-FinkSSU #learning #multi #online
- Online multiclass learning by interclass hypothesis sharing (MF, SSS, YS, SU), pp. 313–320.
- ICML-2006-Garcke
- Regression with the optimised combination technique (JG), pp. 321–328.
- ICML-2006-GeJ #approximate #consistency #multi
- A note on mixtures of experts for multiclass responses: approximation rate and Consistent Bayesian Inference (YG, WJ), pp. 329–335.
- ICML-2006-GehlerHW #adaptation #information retrieval #recognition
- The rate adapting poisson model for information retrieval and object recognition (PVG, AH, MW), pp. 337–344.
- ICML-2006-GeurtsWd #kernel
- Kernelizing the output of tree-based methods (PG, LW, FdB), pp. 345–352.
- ICML-2006-GlobersonR #learning #robust
- Nightmare at test time: robust learning by feature deletion (AG, STR), pp. 353–360.
- ICML-2006-GorurJR #infinity
- A choice model with infinitely many latent features (DG, FJ, CER), pp. 361–368.
- ICML-2006-GravesFGS #classification #network #sequence
- Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks (AG, SF, FJG, JS), pp. 369–376.
- ICML-2006-GreeneC #clustering #documentation #kernel #problem
- Practical solutions to the problem of diagonal dominance in kernel document clustering (DG, PC), pp. 377–384.
- ICML-2006-Haffner #kernel #learning #performance
- Fast transpose methods for kernel learning on sparse data (PH), pp. 385–392.
- ICML-2006-Hanneke #analysis #graph #learning
- An analysis of graph cut size for transductive learning (SH), pp. 393–399.
- ICML-2006-HertzBW #classification #kernel #learning
- Learning a kernel function for classification with small training samples (TH, ABH, DW), pp. 401–408.
- ICML-2006-HolmesJ
- Looping suffix tree-based inference of partially observable hidden state (MPH, CLIJ), pp. 409–416.
- ICML-2006-HoiJZL #classification #image #learning
- Batch mode active learning and its application to medical image classification (SCHH, RJ, JZ, MRL), pp. 417–424.
- ICML-2006-HuangLW #ranking
- Ranking individuals by group comparisons (TKH, CJL, RCW), pp. 425–432.
- ICML-2006-HutchinsonMR #modelling #process
- Hidden process models (RAH, TMM, IR), pp. 433–440.
- ICML-2006-Juba
- Estimating relatedness via data compression (BJ), pp. 441–448.
- ICML-2006-KellerMP #approximate #automation #learning #programming
- Automatic basis function construction for approximate dynamic programming and reinforcement learning (PWK, SM, DP), pp. 449–456.
- ICML-2006-KienzleC #personalisation #recognition
- Personalized handwriting recognition via biased regularization (WK, KC), pp. 457–464.
- ICML-2006-KimMB #analysis #kernel
- Optimal kernel selection in Kernel Fisher discriminant analysis (SJK, AM, SPB), pp. 465–472.
- ICML-2006-KimMSBL #classification #linear
- Pareto optimal linear classification (SJK, AM, SS, SPB, JL), pp. 473–480.
- ICML-2006-KlaasBFDML #performance
- Fast particle smoothing: if I had a million particles (MK, MB, NdF, AD, SM, DL), pp. 481–488.
- ICML-2006-KonidarisB #information management #learning
- Autonomous shaping: knowledge transfer in reinforcement learning (GK, AGB), pp. 489–496.
- ICML-2006-KrauseLG #topic
- Data association for topic intensity tracking (AK, JL, CG), pp. 497–504.
- ICML-2006-KulisSD #kernel #learning #matrix #rank
- Learning low-rank kernel matrices (BK, MAS, ISD), pp. 505–512.
- ICML-2006-LawrenceC #constraints #distance
- Local distance preservation in the GP-LVM through back constraints (NDL, JQC), pp. 513–520.
- ICML-2006-LeSG #knowledge-based
- Simpler knowledge-based support vector machines (QVL, AJS, TG), pp. 521–528.
- ICML-2006-LeeGW #classification #using
- Using query-specific variance estimates to combine Bayesian classifiers (CHL, RG, SW), pp. 529–536.
- ICML-2006-LehmannS #kernel #probability
- A probabilistic model for text kernels (ADL, JST), pp. 537–544.
- ICML-2006-LeordeanuH #approximate #energy #performance
- Efficient MAP approximation for dense energy functions (ML, MH), pp. 545–552.
- ICML-2006-LewisJN #kernel
- Nonstationary kernel combination (DPL, TJ, WSN), pp. 553–560.
- ICML-2006-LiLC #markov #process
- Region-based value iteration for partially observable Markov decision processes (HL, XL, LC), pp. 561–568.
- ICML-2006-Li #clustering #multi
- Multiclass boosting with repartitioning (LL), pp. 569–576.
- ICML-2006-LiM #correlation #modelling #topic
- Pachinko allocation: DAG-structured mixture models of topic correlations (WL, AM), pp. 577–584.
- ICML-2006-LongZWY #clustering #multi #relational
- Spectral clustering for multi-type relational data (BL, Z(Z, XW, PSY), pp. 585–592.
- ICML-2006-LuV #clustering
- Combined central and subspace clustering for computer vision applications (LL, RV), pp. 593–600.
- ICML-2006-MaggioniM #analysis #evaluation #markov #multi #performance #policy #process #using
- Fast direct policy evaluation using multiscale analysis of Markov diffusion processes (MM, SM), pp. 601–608.
- ICML-2006-Martinez-MunozS #order
- Pruning in ordered bagging ensembles (GMM, AS), pp. 609–616.
- ICML-2006-McAuleyCSF #higher-order #image #learning
- Learning high-order MRF priors of color images (JJM, TSC, AJS, MOF), pp. 617–624.
- ICML-2006-Meila
- The uniqueness of a good optimum for K-means (MM), pp. 625–632.
- ICML-2006-Memisevic #kernel
- Kernel information embeddings (RM), pp. 633–640.
- ICML-2006-MoghaddamWA #bound
- Generalized spectral bounds for sparse LDA (BM, YW, SA), pp. 641–648.
- ICML-2006-NaorR #learning
- Learning to impersonate (MN, GNR), pp. 649–656.
- ICML-2006-NarasimhanVS #constraints #latency #markov #modelling #online
- Online decoding of Markov models under latency constraints (MN, PAV, MS), pp. 657–664.
- ICML-2006-NejatiLK #learning #network
- Learning hierarchical task networks by observation (NN, PL, TK), pp. 665–672.
- ICML-2006-NevmyvakaFK #execution #learning
- Reinforcement learning for optimized trade execution (YN, YF, MK), pp. 673–680.
- ICML-2006-PandaCW #bound #concept #detection
- Concept boundary detection for speeding up SVMs (NP, EYC, GW), pp. 681–688.
- ICML-2006-PereiraG #composition
- The support vector decomposition machine (FP, GJG), pp. 689–696.
- ICML-2006-PoupartVHR #learning
- An analytic solution to discrete Bayesian reinforcement learning (PP, NAV, JH, KR), pp. 697–704.
- ICML-2006-RahmaniG #learning #multi #named
- MISSL: multiple-instance semi-supervised learning (RR, SAG), pp. 705–712.
- ICML-2006-RainaNK #learning #using
- Constructing informative priors using transfer learning (RR, AYN, DK), pp. 713–720.
- ICML-2006-RalaivolaDM
- CN = CPCN (LR, FD, CNM), pp. 721–728.
- ICML-2006-RatliffBZ
- Maximum margin planning (NDR, JAB, MZ), pp. 729–736.
- ICML-2006-RavikumarL #estimation #markov #metric #polynomial #programming #random
- Quadratic programming relaxations for metric labeling and Markov random field MAP estimation (PDR, JDL), pp. 737–744.
- ICML-2006-RendersGGPC #categorisation #multi
- Categorization in multiple category systems (JMR, ÉG, CG, FP, GC), pp. 745–752.
- ICML-2006-ReyzinS #classification #complexity #how
- How boosting the margin can also boost classifier complexity (LR, RES), pp. 753–760.
- ICML-2006-RossOZ
- Combining discriminative features to infer complex trajectories (DAR, SO, RSZ), pp. 761–768.
- ICML-2006-RoureM
- Sequential update of ADtrees (JR, AWM), pp. 769–776.
- ICML-2006-RudaryS #modelling #predict #probability
- Predictive linear-Gaussian models of controlled stochastic dynamical systems (MRR, SPS), pp. 777–784.
- ICML-2006-RuckertK #approach #learning #statistics
- A statistical approach to rule learning (UR, SK), pp. 785–792.
- ICML-2006-Sarawagi #modelling #performance #segmentation #sequence
- Efficient inference on sequence segmentation models (SS), pp. 793–800.
- ICML-2006-SenG #learning #markov #network
- Cost-sensitive learning with conditional Markov networks (PS, LG), pp. 801–808.
- ICML-2006-ShengL #algorithm #testing
- Feature value acquisition in testing: a sequential batch test algorithm (VSS, CXL), pp. 809–816.
- ICML-2006-ShivaswamyJ #invariant #permutation
- Permutation invariant SVMs (PKS, TJ), pp. 817–824.
- ICML-2006-SilvaS #learning #metric #modelling
- Bayesian learning of measurement and structural models (RBdAeS, RS), pp. 825–832.
- ICML-2006-SimsekB #performance
- An intrinsic reward mechanism for efficient exploration (ÖS, AGB), pp. 833–840.
- ICML-2006-SindhwaniKC #kernel
- Deterministic annealing for semi-supervised kernel machines (VS, SSK, OC), pp. 841–848.
- ICML-2006-SinghiL #bias #classification #learning #set
- Feature subset selection bias for classification learning (SKS, HL), pp. 849–856.
- ICML-2006-SongE #human-computer #interface #learning
- Classifying EEG for brain-computer interfaces: learning optimal filters for dynamical system features (LS, JE), pp. 857–864.
- ICML-2006-SrebroSR #clustering
- An investigation of computational and informational limits in Gaussian mixture clustering (NS, GS, STR), pp. 865–872.
- ICML-2006-SternHG #game studies #predict #ranking
- Bayesian pattern ranking for move prediction in the game of Go (DHS, RH, TG), pp. 873–880.
- ICML-2006-StrehlLWLL #learning
- PAC model-free reinforcement learning (ALS, LL, EW, JL, MLL), pp. 881–888.
- ICML-2006-StrehlMLH #learning #problem
- Experience-efficient learning in associative bandit problems (ALS, CM, MLL, HH), pp. 889–896.
- ICML-2006-SuZ #classification #network
- Full Bayesian network classifiers (JS, HZ), pp. 897–904.
- ICML-2006-Sugiyama #analysis #reduction
- Local Fisher discriminant analysis for supervised dimensionality reduction (MS), pp. 905–912.
- ICML-2006-SunL
- Iterative RELIEF for feature weighting (YS, JL), pp. 913–920.
- ICML-2006-TangM #multi
- Multiclass reduced-set support vector machines (BT, DM), pp. 921–928.
- ICML-2006-TeoV #array #kernel #performance #string #using
- Fast and space efficient string kernels using suffix arrays (CHT, SVNV), pp. 929–936.
- ICML-2006-TingDS
- Bayesian regression with input noise for high dimensional data (JAT, AD, SS), pp. 937–944.
- ICML-2006-ToussaintS #markov #probability #process
- Probabilistic inference for solving discrete and continuous state Markov Decision Processes (MT, AJS), pp. 945–952.
- ICML-2006-TsudaK #clustering #graph #mining
- Clustering graphs by weighted substructure mining (KT, TK), pp. 953–960.
- ICML-2006-VeeramachaneniOA #detection
- Active sampling for detecting irrelevant features (SV, EO, PA), pp. 961–968.
- ICML-2006-VishwanathanSSM #probability #random
- Accelerated training of conditional random fields with stochastic gradient methods (SVNV, NNS, MWS, KPM), pp. 969–976.
- ICML-2006-Wallach #modelling #topic
- Topic modeling: beyond bag-of-words (HMW), pp. 977–984.
- ICML-2006-WangZ #linear
- Label propagation through linear neighborhoods (FW, CZ), pp. 985–992.
- ICML-2006-WangYL #2d
- Two-dimensional solution path for support vector regression (GW, DYY, FHL), pp. 993–1000.
- ICML-2006-WarmuthLR #algorithm
- Totally corrective boosting algorithms that maximize the margin (MKW, JL, GR), pp. 1001–1008.
- ICML-2006-WestonCSBV
- Inference with the Universum (JW, RC, FHS, LB, VV), pp. 1009–1016.
- ICML-2006-WingateS #kernel #linear #modelling #predict #probability
- Kernel Predictive Linear Gaussian models for nonlinear stochastic dynamical systems (DW, SPS), pp. 1017–1024.
- ICML-2006-WolfeS #predict
- Predictive state representations with options (BW, SPS), pp. 1025–1032.
- ICML-2006-XiKSWR #classification #performance #reduction #using
- Fast time series classification using numerosity reduction (XX, EJK, CRS, LW, CAR), pp. 1033–1040.
- ICML-2006-XiaoSB #reduction
- A duality view of spectral methods for dimensionality reduction (LX, JS, SPB), pp. 1041–1048.
- ICML-2006-XingSJT #multi #process #type inference
- Bayesian multi-population haplotype inference via a hierarchical dirichlet process mixture (EPX, KAS, MIJ, YWT), pp. 1049–1056.
- ICML-2006-XuWSS #learning #predict
- Discriminative unsupervised learning of structured predictors (LX, DFW, FS, DS), pp. 1057–1064.
- ICML-2006-YangFZB #reduction
- Semi-supervised nonlinear dimensionality reduction (XY, HF, HZ, JLB), pp. 1065–1072.
- ICML-2006-YeX #analysis #linear #null #orthogonal
- Null space versus orthogonal linear discriminant analysis (JY, TX), pp. 1073–1080.
- ICML-2006-YuBT #design #learning
- Active learning via transductive experimental design (KY, JB, VT), pp. 1081–1088.
- ICML-2006-YuYTK #collaboration
- Collaborative ordinal regression (SY, KY, VT, HPK), pp. 1089–1096.
- ICML-2006-ZhangK #kernel #matrix #performance
- Block-quantized kernel matrix for fast spectral embedding (KZ, JTK), pp. 1097–1104.
- ICML-2006-ZhengJLNA #debugging #identification #multi #statistics
- Statistical debugging: simultaneous identification of multiple bugs (AXZ, MIJ, BL, MN, AA), pp. 1105–1112.
- ICML-2006-ZhengW #lazy evaluation #performance
- Efficient lazy elimination for averaged one-dependence estimators (FZ, GIW), pp. 1113–1120.

39 ×#learning

14 ×#kernel

13 ×#performance

12 ×#classification

12 ×#multi

11 ×#modelling

10 ×#clustering

10 ×#using

9 ×#predict

7 ×#analysis

14 ×#kernel

13 ×#performance

12 ×#classification

12 ×#multi

11 ×#modelling

10 ×#clustering

10 ×#using

9 ×#predict

7 ×#analysis