Incorrectly labelled training data are frustratingly ubiquitous in both benchmark and specially curated datasets. Such mislabelling clearly adversely affects the performance and generalizability of models trained through supervised learning on the associated datasets. Frameworks for detecting label errors typically require well-trained / well-generalized models; however, at the same time most frameworks rely on training these models on corrupt data, which clearly has the effect of reducing model generalizability and subsequent effectiveness in error detection -- unless a training scheme robust to label errors is employed. We evaluate two novel loss functions, Blurry Loss and Piecewise-zero Loss, that enhance robustness to label errors by de-weighting or disregarding difficult-to-classify samples, which are likely to be erroneous. These loss functions leverage the idea that mislabelled examples are typically more difficult to classify and should contribute less to the learning signal. Comprehensive experiments on a variety of artificially corrupted datasets demonstrate that the proposed loss functions outperform state-of-the-art robust loss functions in nearly all cases, achieving superior F1 scores for error detection. Further analyses through ablation studies offer insights to confirm these loss functions' broad applicability to cases of both uniform and non-uniform corruption, and with different label error detection frameworks. By using these robust loss functions, machine learning practitioners can more effectively identify, prune, or correct errors in their training data.

From Hollywood strikes to digital portraits, AI's potential to steal creatives' work and how to stop it has dominated the tech conversation in 2023. The latest effort to protect artists and their creations is Nightshade, a tool allowing artists to add undetectable pixels into their work that could corrupt an AI's training data, the MIT Technology Review reports. University of Chicago professor Ben Zhao and his team created Nightshade, which is currently being peer reviewed, in an effort to put some of the power back in artists' hands. They tested it on recent Stable Diffusion models and an AI they personally built from scratch. Nightshade essentially works as a poison, altering how a machine-learning model produces content and what that finished product looks like.

With the various terms surrounding Artificial Intelligence (AI) and use cases in business today, it is hard to keep up with all of the new innovation across industries. As AI technology and techniques continue to evolve around us, so do the businesses that use them. Artificial Intelligence can be simply defined in one sentence as the science and engineering of making computers behave in ways that, until recently, we thought required human intelligence. TowardsAI reports, in contrast to Machine Learning, AI is a moving target, and its definition changes as its related technological advancements turn out to be further developed. Machine Learning is one of the ways we expect to achieve AI.