Margin Based Active Learning

By Maria-Florina Balcan, Andrei Z. Broder, Tong Zhang. 2007

412 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

The Unreasonable Effectiveness of Noisy Data for Fine-Grained Recognition (2015, neutral): “This contrasts with common approaches that focus sampling in uncertain regions [28, 1, 35, 14].”
A bound on the label complexity of agnostic active learning (2007, background): “For example, Balcan, Broder & Zhang (Balcan et al., 2007) present an algorithm for linear separa-”
Hierarchical sampling for active learning (2008, background): “Indeed, there are canonical examples in which active learning provably yields exponentially lower label complexity than supervised learning (Cohn et al., 1994; Freund et al., 1997; Dasgupta, 2005; Balcan et al., 2006; Balcan et al., 2007; Castro & Nowak, 2007; Hanneke, 2007; Dasgupta et al., 2007).”
A compression technique for analyzing disagreement-based active learning (2014, neutral): “Speciﬁcally, the following condition of Mammen and Tsybakov (1999) has been studied for several algorithms (see e.g., Balcan, Broder, and Zhang, 2007; Hanneke, 2011; Koltchinskii, 2010; Han-neke, 2012; Hanneke and Yang, 2012; Hanneke, 2014; Beygelzimer, Hsu, Langford, and Zhang, 2010; Hsu, 2010).”
Rates of convergence in active learning (2011, positive): “Subsequently, Balcan, Broder and Zhang [7] proved an analogous positive result for higher-dimensional linear separators, and Castro and Nowak [12] additionally showed a related result for boundary fragment classes (see below); in both cases, the algorithm depends explicitly on the noise parameters.”
Agnostic active learning (2006, background): “6 The main contribution of some of the other papers [23,5] was to show that one can get the desired query complexity in polynomial time and using a polynomial number of unlabeled examples, perhaps also in an online model.”
A theory of transfer learning with applications to active learning (2012, background): “After training on a number of individuals, a learning sys-tem can identify common patterns of speech, such as accents or dialects, each of which requires a slightly different speech recognizer; then, given a new person to train a recognizer for, it can quickly determine the particular dialect from…”
The true sample complexity of active learning (2010, background): “Since C1 trivially has sample complexity 1, and (C2,D) is known to be learnable at an exponential rate (Dasgupta 2005; Balcan et al. 2007; Hanneke 2007a; Dasgupta et al. 2005), combined with Theorem 2, this would imply the result.”
Adversarial Active Learning for Deep Networks: a Margin Based Approach (2018, background): “Deep-Fool attacks In (Balcan et al., 2007), Balcan et al . demonstrated the signiﬁcant beneﬁt of margin-based approaches in reducing human annotations.”
Efficient and Parsimonious Agnostic Active Learning (2015, result): “…the monograph of Hanneke [2014] for an overview of the existing literature, including alternative settings where additional assumptions are placed on the data source (e.g., separability) as is common in other works [Dasgupta, 2005, Balcan et al., 2007, Balcan and Long, 2013]. ar X iv :1 50 6.”
Active and passive learning of linear separators under log-concave distributions (2012, methodology): “…optimal classiﬁer, there exist known parameters α, a > 0 such that, for all w , we have By generalizing Theorem 4 so that it provides a stronger bound for larger margins, and combining the result with the other lemmas of this paper and techniques from (Balcan et al., 2007), we get the following.”
A brief introduction to weakly supervised learning (2018, background): “By assuming Tsybakov noise model, it has been proven that exponential improvement can be obtained for bounded noise [7, 38]; if some special data characteristics, such as multiview structure, can be exploited, exponential improvement can even be achieved for unbounded noise [79].”
The power of localization for efficiently learning linear separators with noise (2013, methodology): “This idea was used in [4] to obtain active learning algorithms with improved label complexity ignoring computational complexity considerations 2 .”
Active Learning as Non-Convex Optimization (2009, methodology): “The algorithm is inspired by the algorithm of Dasgupta et al. (2005) and the offline active learning algorithm of Balcan et al. (2007) (in these papers the threshold changes, the threshold can be less than one, and the updates are different).”
Two faces of active learning (2011, background): “There are now several learning algorithms which, on a variety of canonical examples, provably yield signiﬁcantly lower label complexity than supervised learning [9, 13, 10, 2, 3, 7, 15, 16, 12].”