# Losses over Labels: Weakly Supervised Learning via Direct Loss Construction

arXiv: https://arxiv.org/abs/2212.06921 · HTML: https://mlsanity.com/paper/2212.06921/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- Builds on 33 of its 56 references that are in the map.
- First posted on arXiv 13 Dec 2022.

## Builds on (most cited first, 30 of 33)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Optuna: A Next-generation Hyperparameter Optimization Framework](https://mlsanity.com/paper/1907.10902/) (2019)
- [Certified Adversarial Robustness via Randomized Smoothing](https://mlsanity.com/paper/1902.02918/) (2019)
- [Making Deep Neural Networks Robust to Label Noise: a Loss Correction Approach](https://mlsanity.com/paper/1609.03683/) (2016)
- [Zero-Shot Learning -- A Comprehensive Evaluation of the Good, the Bad and the Ugly](https://mlsanity.com/paper/1707.00600/) (2017)
- [Meta Pseudo Labels](https://mlsanity.com/paper/2003.10580/) (2020)
- [Snorkel: Rapid Training Data Creation with Weak Supervision](https://mlsanity.com/paper/1711.10160/) (2017)
- [Data Programming: Creating Large Training Sets, Quickly](https://mlsanity.com/paper/1605.07723/) (2016)
- [A Survey of Label-noise Representation Learning: Past, Present and Future](https://mlsanity.com/paper/2011.04406/) (2020)
- [Training Complex Models with Multi-Task Weak Supervision](https://mlsanity.com/paper/1810.02840/) (2018)
- [Fast and Three-rious: Speeding Up Weak Supervision with Triplet Methods](https://mlsanity.com/paper/2002.11955/) (2020)
- [WRENCH: A Comprehensive Benchmark for Weak Supervision](https://mlsanity.com/paper/2109.11377/) (2021)
- [Fine-Tuning Pre-trained Language Model with Weak Supervision: A Contrastive-Regularized Self-Training Approach](https://mlsanity.com/paper/2010.07835/) (2020)
- [A Survey on Programmatic Weak Supervision](https://mlsanity.com/paper/2202.05433/) (2022)
- [Learning from Rules Generalizing Labeled Exemplars](https://mlsanity.com/paper/2004.06025/) (2020)
- [Snorkel DryBell: A Case Study in Deploying Weak Supervision at Industrial Scale](https://mlsanity.com/paper/1812.00417/) (2018)
- [Trace Norm Regularised Deep Multi-Task Learning](https://mlsanity.com/paper/1606.04038/) (2016)
- [Learning the Structure of Generative Models without Labeled Data](https://mlsanity.com/paper/1703.00854/) (2017)
- [Self-Training with Weak Supervision](https://mlsanity.com/paper/2104.05514/) (2021)
- [End-to-End Weak Supervision](https://mlsanity.com/paper/2107.02233/) (2021)
- [A Principled Approach for Learning Task Similarity in Multitask Learning](https://mlsanity.com/paper/1903.09109/) (2019)
- [Latent Multi-task Architecture Learning](https://mlsanity.com/paper/1705.08142/) (2017)
- [Universalizing Weak Supervision](https://mlsanity.com/paper/2112.03865/) (2021)
- [DP-SSL: Towards Robust Semi-supervised Learning with A Few Labeled Samples](https://mlsanity.com/paper/2110.13740/) (2021)
- [Curriculum Labeling: Revisiting Pseudo-Labeling for Semi-Supervised Learning](https://mlsanity.com/paper/2001.06001/) (2020)
- [Learning from Multiple Noisy Partial Labelers](https://mlsanity.com/paper/2106.04530/) (2021)
- [Creating Training Sets via Weak Indirect Supervision](https://mlsanity.com/paper/2110.03484/) (2021)
- [Active WeaSuL: Improving Weak Supervision with Active Learning](https://mlsanity.com/paper/2104.14847/) (2021)
- [Constrained Labeling for Weakly Supervised Learning](https://mlsanity.com/paper/2009.07360/) (2020)
- [Optimally Combining Classifiers Using Unlabeled Data](https://mlsanity.com/paper/1503.01811/) (2015)

## Built on by (most cited first, 11 of 11)

- [Theoretical Analysis of Weak-to-Strong Generalization](https://mlsanity.com/paper/2405.16043/) (2024)
- [From Google Gemini to OpenAI Q* (Q-Star): A Survey of Reshaping the Generative Artificial Intelligence (AI) Research Landscape](https://mlsanity.com/paper/2312.10868/) (2023)
- [Predicting the Performance of Black-box LLMs through Follow-up Queries](https://mlsanity.com/paper/2501.01558/) (2025)
- [Learning with Explanation Constraints](https://mlsanity.com/paper/2303.14496/) (2023)
- [Bayesian Neural Networks with Domain Knowledge Priors](https://mlsanity.com/paper/2402.13410/) (2024)
- [Finetuning CLIP to Reason about Pairwise Differences](https://mlsanity.com/paper/2409.09721/) (2024)
- [Weak Supervision Performance Evaluation via Partial Identification](https://mlsanity.com/paper/2312.04601/) (2023)
- [Analyzing Similarity Metrics for Data Selection for Language Model Pretraining](https://mlsanity.com/paper/2502.02494/) (2025)
- [Weakly Supervised Label Learning Flows](https://mlsanity.com/paper/2302.09649/) (2023)
- [Local Boosting for Weakly-Supervised Learning](https://mlsanity.com/paper/2306.02859/) (2023)
- [INSITE: labelling medical images using submodular functions and semi-supervised data programming](https://mlsanity.com/paper/2402.07173/) (2024)

## Abstract (arXiv)

Owing to the prohibitive costs of generating large amounts of labeled data, programmatic weak supervision is a growing paradigm within machine learning. In this setting, users design heuristics that provide noisy labels for subsets of the data. These weak labels are combined (typically via a graphical model) to form pseudolabels, which are then used to train a downstream model. In this work, we question a foundational premise of the typical weakly supervised learning pipeline: given that the heuristic provides all ``label" information, why do we need to generate pseudolabels at all? Instead, we propose to directly transform the heuristics themselves into corresponding loss functions that penalize differences between our model and the heuristic. By constructing losses directly from the heuristics, we can incorporate more information than is used in the standard weakly supervised pipeline, such as how the heuristics make their decisions, which explicitly informs feature selection during training. We call our method Losses over Labels (LoL) as it creates losses directly from heuristics without going through the intermediate step of a label. We show that LoL improves upon existing weak supervision methods on several benchmark text and image classification tasks and further demonstrate that incorporating gradient information leads to better performance on almost every task.