Neural Information Processing Systems http://nips.cc/

The three-dimensional structure of a protein plays a key role in determining its function. Methods like AlphaFold have revolutionized protein structure prediction based only on the amino-acid sequence. However, proteins often appear in multiple different conformations, and it is highly relevant to resolve the full conformational distribution. Single-particle cryo-electron microscopy (cryo EM) is a powerful tool for capturing a large number of images of a given protein, frequently in different conformations (referred to as particles). The images are, however, very noisy projections of the protein, and traditional methods for cryo EM reconstruction are limited to recovering a single, or a few, conformations. In this paper, we introduce cryoSPHERE, a deep learning method that takes as input a nominal protein structure, e.g. from AlphaFold, learns how to divide it into segments, and how to move these as approximately rigid bodies to fit the different conformations present in the cryo EM dataset. This formulation is shown to provide enough constraints to recover meaningful reconstructions of single protein structures. This is illustrated in three examples where we show consistent improvements over the current state-of-the-art for heterogeneous reconstruction.

In recent years, deep learning revolutionized machine learning and its applications, producing results comparable to human experts in several domains, including neuroscience. Each year, hundreds of scientific publications present applications of deep neural networks for biomedical data analysis. Due to the fast growth of the domain, it could be a complicated and extremely time-consuming task for worldwide researchers to have a clear perspective of the most recent and advanced software libraries. This work contributes to clarify the current situation in the domain, outlining the most useful libraries that implement and facilitate deep learning application to neuroscience, allowing scientists to identify the most suitable options for their research or clinical projects. This paper summarizes the main developments in Deep Learning and their relevance to Neuroscience; it then reviews neuroinformatic toolboxes and libraries, collected from the literature and from specific hubs of software projects oriented to neuroscience research. The selected tools are presented in tables detailing key features grouped by domain of application (e.g. data type, neuroscience area, task), model engineering (e.g. programming language, model customization) and technological aspect (e.g. interface, code source). The results show that, among a high number of available software tools, several libraries are standing out in terms of functionalities for neuroscience applications. The aggregation and discussion of this information can help the neuroscience community to devolop their research projects more efficiently and quickly, both by means of readily available tools, and by knowing which modules may be improved, connected or added.

In order to operate in human environments, a robot's semantic perception has to overcome open-world challenges such as novel objects and domain gaps. Autonomous deployment to such environments therefore requires robots to update their knowledge and learn without supervision. We investigate how a robot can autonomously discover novel semantic classes and improve accuracy on known classes when exploring an unknown environment. To this end, we develop a general framework for mapping and clustering that we then use to generate a self-supervised learning signal to update a semantic segmentation model. In particular, we show how clustering parameters can be optimized during deployment and that fusion of multiple observation modalities improves novel object discovery compared to prior work. Models, data, and implementations can be found at github.com/hermannsblum/scim.

One of the ultimate promises of computer vision is to help robotic agents perform active tasks, like delivering packages or doing household chores. However, the conventional approach to solving "vision" is to define a set of offline recognition problems (e.g. object detection) and solve those first. This approach faces a challenge from the recent rise of Deep Reinforcement Learning frameworks that learn active tasks from scratch using images as input. This poses a set of fundamental questions: what is the role of computer vision if everything can be learned from scratch? Could intermediate vision tasks actually be useful for performing arbitrary downstream active tasks? We show that proper use of mid-level perception confers significant advantages over training from scratch. We implement a perception module as a set of mid-level visual representations and demonstrate that learning active tasks with mid-level features is significantly more sample-efficient than scratch and able to generalize in situations where the from-scratch approach fails. However, we show that realizing these gains requires careful selection of the particular mid-level features for each downstream task. Finally, we put forth a simple and efficient perception module based on the results of our study, which can be adopted as a rather generic perception module for active frameworks.

Do visual tasks have a relationship, or are they unrelated? For instance, could having surface normals simplify estimating the depth of an image? Intuition answers these questions positively, implying existence of a structure among visual tasks. Knowing this structure has notable values; it is the concept underlying transfer learning and provides a principled way for identifying redundancies across tasks, e.g., to seamlessly reuse supervision among related tasks or solve many tasks in one system without piling up the complexity. We proposes a fully computational approach for modeling the structure of space of visual tasks. This is done via finding (first and higher-order) transfer learning dependencies across a dictionary of twenty six 2D, 2.5D, 3D, and semantic tasks in a latent space. The product is a computational taxonomic map for task transfer learning. We study the consequences of this structure, e.g. nontrivial emerged relationships, and exploit them to reduce the demand for labeled data. For example, we show that the total number of labeled datapoints needed for solving a set of 10 tasks can be reduced by roughly 2/3 (compared to training independently) while keeping the performance nearly the same. We provide a set of tools for computing and probing this taxonomical structure including a solver that users can employ to devise efficient supervision policies for their use cases.