A deep neural network (DNN) has achieved great success in many machine learning tasks by virtue of its high expressive power. However, its prediction can be easily biased to undesirable features, which are not essential for solving the target task and are even imperceptible to a human, thereby resulting in poor generalization. Leveraging plenty of undesirable features in out-of-distribution (OOD) examples has emerged as a potential solution for de-biasing such features, and a recent study shows that softmax-level calibration of OOD examples can successfully remove the contribution of undesirable features to the last fully-connected layer of a classifier. However, its applicability is confined to the classification task, and its impact on a DNN feature extractor is not properly investigated. In this paper, we propose TAUFE, a novel regularizer that deactivates many undesirable features using OOD examples in the feature extraction layer and thus removes the dependency on the task-specific softmax layer. To show the task-agnostic nature of TAUFE,we rigorously validate its performance on three tasks, classification, regression, and a mix of them, on CIFAR-10, CIFAR-100, ImageNet, CUB200, and CAR datasets. The results demonstrate that TAUFE consistently outperforms the state-of-the-art method as well as the baselines without regularization.

CLIPDraw is an algorithm that synthesizes novel drawings from natural language input. It does not require any additional training; rather, a pre-trained CLIP language-image encoder is used as a metric for maximizing similarity between the given description and a generated drawing. Crucially, CLIPDraw operates over vector strokes rather than pixel images, which biases drawings towards simpler human-recognizable shapes. Results compare CLIPDraw with other synthesisthrough-optimization methods, as well as highlight various interesting behaviors of CLIPDraw, such as satisfying ambiguous text in multiple ways, reliably producing drawings in diverse styles, and scaling from simple to complex visual representations as stroke count increases.

Recently, large-scale pre-trained Vision and Language (VL) models have set a new state-of-the-art (SOTA) in zero-shot visual classification enabling open-vocabulary recognition of potentially unlimited set of categories defined as simple language prompts. However, despite these great advances, the performance of these zeroshot classifiers still falls short of the results of dedicated (closed category set) classifiers trained with supervised fine-tuning. In this paper we show, for the first time, how to reduce this gap without any labels and without any paired VL data, using an unlabeled image collection and a set of texts auto-generated using a Large Language Model (LLM) describing the categories of interest and effectively substituting labeled visual instances of those categories. Using our label-free approach, we are able to attain significant performance improvements over the zero-shot performance of the base VL model and other contemporary methods and baselines on a wide variety of datasets, demonstrating absolute improvement of up to 11.7% (3.8% on average) in the label-free setting. Moreover, despite our approach being label-free, we observe 1.3% average gains over leading few-shot prompting baselines that do use 5-shot supervision.

Cutting-plane methods have enabled remarkable successes in integer programming over the last few decades. State-of-the-art solvers integrate a myriad of cutting-plane techniques to speed up the underlying tree-search algorithm used to find optimal solutions. In this paper we provide sample complexity bounds for cut-selection in branch-and-cut (B&C). Given a training set of integer programs sampled from an application-specific input distribution and a family of cut selection policies, these guarantees bound the number of samples sufficient to ensure that using any policy in the family, the size of the tree B&C builds on average over the training set is close to the expected size of the tree B&C builds. We first bound the sample complexity of learning cutting planes from the canonical family of Chvátal-Gomory cuts. Our bounds handle any number of waves of any number of cuts and are fine tuned to the magnitudes of the constraint coefficients. Next, we prove sample complexity bounds for more sophisticated cut selection policies that use a combination of scoring rules to choose from a family of cuts. Finally, beyond the realm of cutting planes for integer programming, we develop a general abstraction of tree search that captures key components such as node selection and variable selection. For this abstraction, we bound the sample complexity of learning a good policy for building the search tree.

For a vector x 2 Rd and H [d], we denote vH to denote the vector that is equal to v on i 2 H, and zero otherwise. For a real-valued random variable X and m 2 N, we use kXkLm to denote (E|X|m)1/m. For a set S Rd and a function f, we also define the set function notation f(S) as {f(x)|x 2 S}. A.1 Finding a stable subset from a stable weighted subset For a set S on npoints, we define n, as the set of weights w 2 Rn such that wi 2 [0,1/((1)n]for all i 2 [n]and P i wi =1 . For a fixed vector µ 2 Rd that will be clear from context, a set of npoints S = {x1,...,x n}, and weights w 2 n, over S, we use w to denote P i wi(xi µ)(xi µ)>. The goal of this section is to show Proposition A.1, which states that if we have a weight w over S such that w (with respect to some vector µ) has bounded Xk norm proportional to 2 for some > 0, then there must exists some large subset S0 S that is stable with respect to µ and . Let S be a set of n points in Rd. Suppose that there exists a w 2 n, such that k wkXk B 2 for some vector µ. Then there exists a subset S0 S such that (i)|S0| (1 2)n and (ii) S0 is (,,k)-stable with respect to µ and, where = O( p B +1). Observe that k wkXk B 2 implies k w 2IkXk (B +1) 2 by the triangle inequality. In order to show Proposition A.1, we show Lemma A.2, which is a weakening of Proposition A.1 where we additionally assume that µw = P i wixi is close to µ, where µ is the vector we use to define w as well as the vector that we want to find a large sample subset S0 to be stable with respect to. To use Lemma A.2, we additionally show Proposition A.4, which states that k wkXk B 2 is enough to imply that µw is close to µ.