VC bounds on the cardinality of nearly orthogonal function classes
نویسندگان
چکیده
We bound the number of nearly orthogonal vectors with fixed VCdimension over {−1, 1}n. Our bounds are of interest in machine learning and empirical process theory and improve previous bounds by Haussler. The bounds are based on a simple projection argument and the generalize to other product spaces. Along the way we derive tight bounds on the sum of binomial coefficients in terms of the entropy function.
منابع مشابه
Lower bounds on the signed (total) $k$-domination number
Let $G$ be a graph with vertex set $V(G)$. For any integer $kge 1$, a signed (total) $k$-dominating functionis a function $f: V(G) rightarrow { -1, 1}$ satisfying $sum_{xin N[v]}f(x)ge k$ ($sum_{xin N(v)}f(x)ge k$)for every $vin V(G)$, where $N(v)$ is the neighborhood of $v$ and $N[v]=N(v)cup{v}$. The minimum of the values$sum_{vin V(G)}f(v)$, taken over all signed (total) $k$-dominating functi...
متن کاملGeneralization Bounds for Connected Function Classes
We derive an improvement to the Vapnik-Chervonenkis bounds on generalization error which applies to many commonly used function classes, including feedforward neural networks. The VC analysis simply counts the number of dichotomies induced by a function class on a set of input vectors. We can achieve a richer description of the exibility of a function class by taking into account the diversity ...
متن کاملComplexity of VC-classes of sequences with long repetitive runs
The Vapnik-Chervonenkis (VC) dimension (also known as the trace number) and the Sauer-Shelah lemma have found applications in numerous areas including set theory, combinatorial geometry, graph theory and statistical learning theory. Estimation of the complexity of discrete structures associated with the search space of algorithms often amounts to estimating the cardinality of a simpler class wh...
متن کاملVC-Dimensions of Random Function Classes
For any class of binary functions on [n] = {1, . . . , n} a classical result by Sauer states a sufficient condition for its VC-dimension to be at least d: its cardinality should be at least O(n). A necessary condition is that its cardinality be at least 2 (which is O(1) with respect to n). How does the size of a ‘typical’ class of VC-dimension d compare to these two extreme thresholds ? To answ...
متن کاملOn the VC-dimension and Boolean functions with long runs
The Vapnik-Chervonenkis (VC) dimension and the Sauer-Shelah lemma have found applications in numerous areas including set theory, combinatorial geometry, graph theory and statistical learning theory. Estimation of the complexity of discrete structures associated with the search space of algorithms often amounts to estimating the cardinality of a simpler class which is effectively induced by som...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید
ثبت ناماگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید
ورودعنوان ژورنال:
- Discrete Mathematics
دوره 312 شماره
صفحات -
تاریخ انتشار 2012