CSDL Home IEEE Transactions on Pattern Analysis & Machine Intelligence 2012 vol.34 Issue No.06 - June
Issue No.06 - June (2012 vol.34)
K. P. Bennett , Depts. of Math. Sci. & Comput. Sci., Rensselaer Polytech. Inst., Troy, NY, USA
We present a bundle algorithm for multiple-instance classification and ranking. These frameworks yield improved models on many problems possessing special structure. Multiple-instance loss functions are typically nonsmooth and nonconvex, and current algorithms convert these to smooth nonconvex optimization problems that are solved iteratively. Inspired by the latest linear-time subgradient-based methods for support vector machines, we optimize the objective directly using a nonconvex bundle method. Computational results show this method is linearly scalable, while not sacrificing generalization accuracy, permitting modeling on new and larger data sets in computational chemistry and other applications. This new implementation facilitates modeling with kernels.
pattern classification, convex programming, gradient methods, learning (artificial intelligence), computational chemistry, bundle algorithm, multiple-instance learning, multiple-instance classification, multiple-instance ranking, multiple-instance loss functions, smooth nonconvex optimization problems, linear-time subgradient-based methods, support vector machines, nonconvex bundle method, Kernel, Compounds, Microwave integrated circuits, Drugs, Computational modeling, Support vector machines, Optimization, medicine and science., Artificial intelligence, machine learning, nonsmooth optimization, bundle methods, multiple-instance learning, ranking
K. P. Bennett, "Fast Bundle Algorithm for Multiple-Instance Learning", IEEE Transactions on Pattern Analysis & Machine Intelligence, vol.34, no. 6, pp. 1068-1079, June 2012, doi:10.1109/TPAMI.2011.194