EXACT: How to Train Your Accuracy
Classification tasks are usually evaluated in terms of accuracy. However, accuracy is discontinuous and cannot be directly optimized using gradient ascent. Popular methods minimize cross-entropy, hinge loss, or other surrogate losses, which can lead to suboptimal results. In this paper, we propose a new optimization framework by introducing stochasticity to a model's output and optimizing expected accuracy, i.e. accuracy of the stochastic model. Extensive experiments on linear models and deep image classification show that the proposed optimization method is a powerful alternative to widely used classification losses.
Code (2)
Tasks
General Classificationimage-classificationImage ClassificationSimilar Papers 제목 키워드 기반
PushupBench: Your VLM is not good at counting pushups
Large vision-language models (VLMs) can recognize \textit{what} happens in video but fail to count \textit{how many} times. We introduce \textbf{PushupBench}, 446 long-form clips (avg. 36.7s) for evaluating repetition co…
Slim attention: cut your context memory in half without loss of accuracy -- K-cache is all you need for MHA
Slim attention shrinks the context memory size by 2x for transformer models with MHA (multi-head attention), which can speed up inference by up to 2x for large context windows. Slim attention is an exact, mathematically …
AllDecoderHow to Turn Your Knowledge Graph Embeddings into Generative Models
Some of the most successful knowledge graph embedding (KGE) models for link prediction -- CP, RESCAL, TuckER, ComplEx -- can be interpreted as energy-based models. Under this perspective they are not amenable for exact m…
Graph EmbeddingKnowledge Graph EmbeddingKnowledge Graph EmbeddingsLink PredictionCut your Losses with Squentropy
Nearly all practical neural models for classification are trained using cross-entropy loss. Yet this ubiquitous choice is supported by little theoretical or empirical evidence. Recent work (Hui & Belkin, 2020) suggests t…
ClassificationMulti-class ClassificationAutomated detection of dark patterns in cookie banners: how to do it poorly and why it is hard to do it any other way
Cookie banners, the pop ups that appear to collect your consent for data collection, are a tempting ground for dark patterns. Dark patterns are design elements that are used to influence the user's choice towards an opti…
BIG-bench Machine LearningFeature Engineering