اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدGeneralized Ambiguity Decomposition for Understanding Ensemble Diversity
51
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Diversity or complementarity of experts in ensemble pattern recognition and information processing systems is widely-observed by researchers to be crucial for achieving performance improvement upon fusion. Understanding this link between ensemble diversity and fusion performance is thus an important research question. However, prior works have theoretically characterized ensemble diversity and have linked it with ensemble performance in very restricted settings. We present a generalized ambiguity decomposition (GAD) theorem as a broad framework for answering these questions. The GAD theorem applies to a generic convex ensemble of experts for any arbitrary twice-differentiable loss function. It shows that the ensemble performance approximately decomposes into a difference of the average expert performance and the diversity of the ensemble. It thus provides a theoretical explanation for the empirically-observed benefit of fusing outputs from diverse classifiers and regressors. It also provides a loss function-dependent, ensemble-dependent, and data-dependent definition of diversity. We present extensions of this decomposition to common regression and classification loss functions, and report a simulation-based analysis of the diversity term and the accuracy of the decomposition. We finally present experiments on standard pattern recognition data sets which indicate the accuracy of the decomposition for real-world classification and regression problems.
قيم البحث
اقرأ أيضاً
In this paper we propose a generalization of deep neural networks called deep function machines (DFMs). DFMs act on vector spaces of arbitrary (possibly infinite) dimension and we show that a family of DFMs are invariant to the dimension of input dat
a; that is, the parameterization of the model does not directly hinge on the quality of the input (eg. high resolution images). Using this generalization we provide a new theory of universal approximation of bounded non-linear operators between function spaces. We then suggest that DFMs provide an expressive framework for designing new neural network layer types with topological considerations in mind. Finally, we introduce a novel architecture, RippLeNet, for resolution invariant computer vision, which empirically achieves state of the art invariance.
In this paper we study the training dynamics for gradient flow on over-parametrized tensor decomposition problems. Empirically, such training process often first fits larger components and then discovers smaller components, which is similar to a tens
or deflation process that is commonly used in tensor decomposition algorithms. We prove that for orthogonally decomposable tensor, a slightly modified version of gradient flow would follow a tensor deflation process and recover all the tensor components. Our proof suggests that for orthogonal tensors, gradient flow dynamics works similarly as greedy low-rank learning in the matrix setting, which is a first step towards understanding the implicit regularization effect of over-parametrized models for low-rank tensors.
Classical learning theory suggests that the optimal generalization performance of a machine learning model should occur at an intermediate model complexity, with simpler models exhibiting high bias and more complex models exhibiting high variance of
the predictive function. However, such a simple trade-off does not adequately describe deep learning models that simultaneously attain low bias and variance in the heavily overparameterized regime. A primary obstacle in explaining this behavior is that deep learning algorithms typically involve multiple sources of randomness whose individual contributions are not visible in the total variance. To enable fine-grained analysis, we describe an interpretable, symmetric decomposition of the variance into terms associated with the randomness from sampling, initialization, and the labels. Moreover, we compute the high-dimensional asymptotic behavior of this decomposition for random feature kernel regression, and analyze the strikingly rich phenomenology that arises. We find that the bias decreases monotonically with the network width, but the variance terms exhibit non-monotonic behavior and can diverge at the interpolation boundary, even in the absence of label noise. The divergence is caused by the emph{interaction} between sampling and initialization and can therefore be eliminated by marginalizing over samples (i.e. bagging) emph{or} over the initial parameters (i.e. ensemble learning).
Modern biomedical studies often collect multiple types of high-dimensional data on a common set of objects. A popular model for the joint analysis of multi-type datasets decomposes each data matrix into a low-rank common-variation matrix generated by
latent factors shared across all datasets, a low-rank distinctive-variation matrix corresponding to each dataset, and an additive noise matrix. We propose decomposition-based generalized canonical correlation analysis (D-GCCA), a novel decomposition method that appropriately defines those matrices on the L2 space of random variables, whereas most existing methods are developed on its approximation, the Euclidean dot product space. Moreover to well calibrate common latent factors, we impose a desirable orthogonality constraint on distinctive latent factors. Existing methods inadequately consider such orthogonality and can thus suffer from substantial loss of undetected common variation. Our D-GCCA takes one step further than GCCA by separating common and distinctive variations among canonical variables, and enjoys an appealing interpretation from the perspective of principal component analysis. Consistent estimators of our common-variation and distinctive-variation matrices are established with good finite-sample numerical performance, and have closed-form expressions leading to efficient computation especially for large-scale datasets. The superiority of D-GCCA over state-of-the-art methods is also corroborated in simulations and real-world data examples.
Ensembles of models have been empirically shown to improve predictive performance and to yield robust measures of uncertainty. However, they are expensive in computation and memory. Therefore, recent research has focused on distilling ensembles into
a single compact model, reducing the computational and memory burden of the ensemble while trying to preserve its predictive behavior. Most existing distillation formulations summarize the ensemble by capturing its average predictions. As a result, the diversity of the ensemble predictions, stemming from each member, is lost. Thus, the distilled model cannot provide a measure of uncertainty comparable to that of the original ensemble. To retain more faithfully the diversity of the ensemble, we propose a distillation method based on a single multi-headed neural network, which we refer to as Hydra. The shared body network learns a joint feature representation that enables each head to capture the predictive behavior of each ensemble member. We demonstrate that with a slight increase in parameter count, Hydra improves distillation performance on classification and regression settings while capturing the uncertainty behavior of the original ensemble over both in-domain and out-of-distribution tasks.
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
