Deep metric attention learning for skin lesion classification in dermoscopy images

Currently, convolutional neural networks (CNNs) have made remarkable achievements in skin lesion classification because of their end-to-end feature representation abilities. However, precise skin lesion classification is still challenging because of the following three issues: (1) insufficient training samples, (2) inter-class similarities and intra-class variations, and (3) lack of the ability to focus on discriminative skin lesion parts. To address these issues, we propose a deep metric attention learning CNN (DeMAL-CNN) for skin lesion classification. In DeMAL-CNN, a triplet-based network (TPN) is first designed based on deep metric learning, which consists of three weight-shared embedding extraction networks. TPN adopts a triplet of samples as input and uses the triplet loss to optimize the embeddings, which can not only increase the number of training samples, but also learn the embeddings robust to inter-class similarities and intra-class variations. In addition, a mixed attention mechanism considering both the spatial-wise and channel-wise attention information is designed and integrated into the construction of each embedding extraction network, which can further strengthen the skin lesion localization ability of DeMAL-CNN. After extracting the embeddings, three weight-shared classification layers are used to generate the final predictions. In the training procedure, we combine the triplet loss with the classification loss as a hybrid loss to train DeMAL-CNN. We compare DeMAL-CNN with the baseline method, attention methods, advanced challenge methods, and state-of-the-art skin lesion classification methods on the ISIC 2016 and ISIC 2017 datasets, and test its generalization ability on the PH2 dataset. The results demonstrate its effectiveness.

An approach to partial occlusion using deep metric learning

<span>The human face can be used as an identification and authentication tool in biometric systems. Face recognition in forensics is a challenging task due to the presence of partial occlusion features like wearing a hat, sunglasses, scarf, and beard. In forensics, criminal identification having partial occlusion features is the most difficult task to perform. In this paper, a combination of the histogram of gradients (HOG) with Euclidean distance is proposed. Deep metric learning is the process of measuring the similarity between the samples using optimal distance metrics for learning tasks. In the proposed system, a deep metric learning technique like HOG is used to generate a 128d real feature vector. Euclidean distance is then applied between the feature vectors and a tolerance threshold is set to decide whether it is a match or mismatch. Experiments are carried out on disguised faces in the wild (DFW) dataset collected from IIIT Delhi which consists of 1000 subjects in which 600 subjects were used for testing and the remaining 400 subjects were used for training purposes. The proposed system provides a recognition accuracy of 89.8% and it outperforms compared with other existing methods.</span>

Deep metric learning improves lab of origin prediction of genetically engineered plasmids

Genome engineering is undergoing unprecedented development and is now becoming widely available. To ensure responsible biotechnology innovation and to reduce misuse of engineered DNA sequences, it is vital to develop tools to identify the lab-of-origin of engineered plasmids. Genetic engineering attribution (GEA), the ability to make sequence-lab associations, would supportforensic experts in this process. Here, we propose a method, based on metric learning, that ranks the most likely labs-of-origin whilstsimultaneously generating embeddings for plasmid sequences and labs. These embeddings can be used to perform various downstreamtasks, such as clustering DNA sequences and labs, as well as using them as features in machine learning models. Our approach employsa circular shift augmentation approach and is able to correctly rank the lab-of-origin90%of the time within its top 10 predictions -outperforming all current state-of-the-art approaches. We also demonstrate that we can perform few-shot-learning and obtain76%top-10 accuracy using only10%of the sequences. This means, we outperform the previous CNN approach using only one-tenth of the data. We also demonstrate that we are able to extract key signatures in plasmid sequences for particular labs, allowing for an interpretable examination of the model’s outputs.CCS Concepts: Information systems→Similarity measures; Learning to rank.

MapCell: Learning a Comparative Cell Type Distance Metric With Siamese Neural Nets With Applications Toward Cell-Type Identification Across Experimental Datasets

Large collections of annotated single-cell RNA sequencing (scRNA-seq) experiments are being generated across different organs, conditions and organisms on different platforms. The diversity, volume and complexity of this aggregated data requires new analysis techniques to extract actionable knowledge. Fundamental to most analysis are key abilities such as: identification of similar cells across different experiments and transferring annotations from an annotated dataset to an unannotated one. There have been many strategies explored in achieving these goals, and they focuses primarily on aligning and re-clustering datasets of interest. In this work, we are interested in exploring the applicability of deep metric learning methods as a form of distance function to capture similarity between cells and facilitate the transfer of cell type annotation for similar cells across different experiments. Toward this aim, we developed MapCell, a few-shot training approach using Siamese Neural Networks (SNNs) to learn a generalizable distance metric that can differentiate between single cell types. Requiring only a small training set, we demonstrated that SNN derived distance metric can perform accurate transfer of annotation across different scRNA-seq platforms, batches, species and also aid in flagging novel cell types.

Function shaping in deep learning

This work describes the importance of loss functions and related methods for deep reinforcement learning and deep metric learning. A novel MDQN loss function outperformed DDQN loss function in PLE computer game environments, and a novel Exponential Triplet loss function outperformed the Triplet loss function in the face re-identification task with VGGFace2 dataset reaching 85,7 % accuracy using zero-shot setting. This work also presents a novel UNet-RNN-Skip model to improve the performance of the value function for path planning tasks.