Dear AI & ML Enthusiasts,
This time we have two exciting speakers.
Axel Dittmann, Microsoft Deutschland GmbH
How to use Azure Machine Learning service to manage the lifecycle of your models. Azure Machine Learning uses a Machine Learning Operations (MLOps) approach, which improves the quality and consistency of your machine learning solutions.
Learning Global Pooling Operators in Deep Neural Networks for Image
Lukas Spranger, Siemens AG
Like other fields of computer vision, image retrieval has been
revolutionized by deep learning in recent years. Convolutional neural
networks are now the tool of choice for computing feature representations of images. Many successful architectures employ global pooling layers to aggregate feature maps to a compact image representation. Using the neural network training procedure based on backpropagation and gradient descent methods, we can learn the global pooling operation from the training data.
We review existing approaches to learned pooling and propose two new
layers: A learnable, extended variant of LSE pooling and the generalized
max pooling layer based on an aggregation function from classical computer vision.
We evaluate the learnable pooling layers on datasets for writer
identification and person re-identification. Our experiments show that
learned global pooling can improve performance of image retrieval networks compared to the average pooling baseline for both tasks. For writer identification, our generalized max pooling layer outperforms all other tested pooling layers. Our learnable LSE pooling performs better than global average pooling and yields the best rank-1 score in our experiments on the Market-1501 dataset.
Looking forward to seeing you all on October 8th!
We are grateful to codemanufaktur for hosting us!
*************** IMPORTANT ***************
Photography / filming / audio recording is taking place during the event for documentation purposes. The photographs and recordings will likely be published for the purpose of reporting on the event on all communication channels of the event organizers. If you do not wish this, please inform the organizer on site before the start of the event.