Are you an EPFL student looking for a semester project?

Work with us on data science and visualisation projects, and deploy your project as an app on top of GraphSearch.

Publication

Using Animal Motion Capture to Learn Neural Representations

Semih Günel
EPFL, 2022
EPFL thesis

Abstract

Understanding behavior from neural activity is a fundamental goal in neuroscience. It has practical applications in building robust brain-machine interfaces, human-computer interaction, and assisting patients with neurological disabilities. Despite the ever-growing demand driven by the applications, neural action decoding methods still require laborious manual annotations. Furthermore, a different model must be trained for each new individual, requiring even more annotation and overwhelming the resources of most applications. This thesis proposes a self-supervised way to tackle the unsupervised domain adaptation for the neural action decoding problem. Our method does not require manual annotations and can generalize across individuals. To do that, we propose to use animal motion capture to guide the encoding of neural action representations together with a set of neural and behavioral augmentations. We, therefore, divide the thesis into two parts. For the first part, we propose multiple novel ways to learn robust markerless motion capture systems for animals. In the second part, we then use motion capture to learn an unsupervised neural action decoding algorithm.To start, we present a novel unsupervised markerless 2D pose estimation system. Extracting pose information from laboratory animals requires significant annotation labor. Our model requires only unrealistic renderings of the target animal, without any manual pose labeling. While deep learning models cannot be trained on unrealistic renderings of laboratory animals due to the domain gap, we propose a deformation-based generative network transforming the renderings into realistic images while at the same time keeping the pose information intact. Our model enables us to train a pose estimation network on generated but realistic images without manual annotations.Second, we propose a markerless self-calibrating motion capture system on tethered animals. Although extracting a 2D pose is enough for some applications, 2D pose information is incomplete due to information loss from the camera projection. On the other hand, triangulating 3D poses from multiple images requires camera calibration. This tedious process is not ideal for many laboratory setups, for example, a small model organism \textit{Drosophila}. We model a self-calibrating system that uses the animal as the calibration pattern. To make the system even more reliable, we use the shape and skeleton of the animal as the 3D prior, therefore fixing the inference mistakes of the deep learning based methods.Third, we propose a monocular 3D motion capture algorithm for laboratory animals. Although 3D pose information provides all the necessary information about the movement, multi-camera systems require expensive hardware and multi-view synchronization. We propose a monocular system that can still recover 3D pose without requiring multiple views. We test our model on numerous lab animals, including flies, mice, rats, and macaques. We show that our monocular motion capture system performs reliably and on-par with multi-camera motion capture systems.Lastly, we propose a novel unsupervised domain adaptation method for neural decoding of actions, which uses the previously mentioned animal markerless motion capture systems for domain adaptation. We show that our neural action decoding algorithm can successfully generalize across individuals. Furthermore, our method can provide better neural acti

Official source

https://infoscience.epfl.ch/record/293821?ln=en

About this result

This page is automatically generated and may contain information that is not correct, complete, up-to-date, or relevant to your search query. The same applies to every other page on this website. Please make sure to verify the information with EPFL's official sources.

Related concepts

Related publications

Semih Günel
EPFL, 2022
EPFL thesis

Abstract

Official source

https://infoscience.epfl.ch/record/293821?ln=en

About this result

Related concepts (16)

Related concepts

Related publications (122)

Related publications

Theory of representation learning in cortical neural networks

Carlos Stein Naves de Brito

Our brain continuously self-organizes to construct and maintain an internal representation of the world based on the information arriving through sensory stimuli. Remarkably, cortical areas related to different sensory modalities appear to share the same functional unit, the neuron, and develop through the same learning mechanism, synaptic plasticity. It motivates the conjecture of a unifying theory to explain cortical representational learning across sensory modalities. In this thesis we present theories and computational models of learning and optimization in neural networks, postulating functional properties of synaptic plasticity that support the apparent universal learning capacity of cortical networks. In the past decades, a variety of theories and models have been proposed to describe receptive field formation in sensory areas. They include normative models such as sparse coding, and bottom-up models such as spike-timing dependent plasticity. We bring together candidate explanations by demonstrating that in fact a single principle is sufficient to explain receptive field development. First, we show that many representative models of sensory development are in fact implementing variations of a common principle: nonlinear Hebbian learning. Second, we reveal that nonlinear Hebbian learning is sufficient for receptive field formation through sensory inputs. A surprising result is that our findings are independent of specific details, and allow for robust predictions of the learned receptive fields. Thus nonlinear Hebbian learning and natural statistics can account for many aspects of receptive field formation across models and sensory modalities. The Hebbian learning theory substantiates that synaptic plasticity can be interpreted as an optimization procedure, implementing stochastic gradient descent. In stochastic gradient descent inputs arrive sequentially, as in sensory streams. However, individual data samples have very little information about the correct learning signal, and it becomes a fundamental problem to know how many samples are required for reliable synaptic changes. Through estimation theory, we develop a novel adaptive learning rate model, that adapts the magnitude of synaptic changes based on the statistics of the learning signal, enabling an optimal use of data samples. Our model has a simple implementation and demonstrates improved learning speed, making this a promising candidate for large artificial neural network applications. The model also makes predictions on how cortical plasticity may modulate synaptic plasticity for optimal learning. The optimal sampling size for reliable learning allows us to estimate optimal learning times for a given model. We apply this theory to derive analytical bounds on times for the optimization of synaptic connections. First, we show this optimization problem to have exponentially many saddle-nodes, which lead to small gradients and slow learning. Second, we show that the number of input synapses to a neuron modulates the magnitude of the initial gradient, determining the duration of learning. Our final result reveals that the learning duration increases supra-linearly with the number of synapses, suggesting an effective limit on synaptic connections and receptive field sizes in developing neural networks.

EPFL2016

The way our brain learns to disentangle complex signals into unambiguous concepts is fascinating but remains largely unknown. There is evidence, however, that hierarchical neural representations play a key role in the cortex. This thesis investigates biologically plausible models of unsupervised learning of hierarchical representations as found in the brain and modern computer vision models. We use computational modeling to address three main questions at the intersection of artificial intelligence (AI) and computational neuroscience.The first question is: What are useful neural representations and when are deep hierarchical representations needed? We approach this point with a systematic study of biologically plausible unsupervised feature learning in a shallow 2-layer networks on digit (MNIST) and object (CIFAR10) classification. Surprisingly, random features support high performance, especially for large hidden layers. When combined with localized receptive fields, random feature networks approach the performance of supervised backpropagation on MNIST, but not on CIFAR10. We suggest that future models of biologically plausible learning should outperform such random feature benchmarks on MNIST, or that such models should be evaluated in different ways.The second question is: How can hierarchical representations be learned with mechanisms supported by neuroscientific evidence? We cover this question by proposing a unifying Hebbian model, inspired by common models of V1 simple and complex cells based on unsupervised sparse coding and temporal invariance learning. In shallow 2-layer networks, our model reproduces learning of simple and complex cell receptive fields, as found in V1. In deeper networks, we stack multiple layers of Hebbian learning but find that it does not yield hierarchical representations of increasing usefulness. From this, we hypothesise that standard Hebbian rules are too constrained to build increasingly useful representations, as observed in higher areas of the visual cortex or deep artificial neural networks.The third question is: Can AI inspire learning models that build deep representations and are still biologically plausible? We address this question by proposing a learning rule that takes inspiration from neuroscience and recent advances in self-supervised deep learning. The proposed rule is Hebbian, i.e. only depends on pre- and post-synaptic neuronal activity, but includes additional local factors, namely predictive dendritic input and widely broadcasted modulation factors. Algorithmically, this rule applies self-supervised contrastive predictive learning to a causal, biological setting using saccades. We find that networks trained with this generalised Hebbian rule build deep hierarchical representations of images, speech and video.We see our modeling as a potential starting point for both, new hypotheses, that can be tested experimentally, and novel AI models that could benefit from added biological realism.

EPFL2021

Understanding Deep Neural Function Approximation in Reinforcement Learning via ϵ-Greedy Exploration

Volkan Cevher, Luca Viano

This paper provides a theoretical study of deep neural function approximation in reinforcement learning (RL) with the ϵ-greedy exploration under the online setting. This problem setting is motivated by the successful deep Q-networks (DQN) framework that falls in this regime. In this work, we provide an initial attempt on theoretical understanding deep RL from the perspective of function class and neural networks architectures (e.g., width and depth) beyond the ``linear'' regime. To be specific, we focus on the value based algorithm with the ϵ-greedy exploration via deep (and two-layer) neural networks endowed by Besov (and Barron) function spaces, respectively, which aims at approximating an α-smooth Q-function in a d-dimensional feature space. We prove that, with T episodes, scaling the width

m = \widetilde{\mathcal{O}}(T^{\frac{d}{2\alpha + d}})

and the depth

L=\mathcal{O}(\log T)

of the neural network for deep RL is sufficient for learning with sublinear regret in Besov spaces. Moreover, for a two layer neural network endowed by the Barron space, scaling the width

\Omega(\sqrt{T})

is sufficient. To achieve this, the key issue in our analysis is how to estimate the temporal difference error under deep neural function approximation as the ϵ-greedy exploration is not enough to ensure "optimism". Our analysis reformulates the temporal difference error in an

L^2(\mathrm{d}\mu)

-integrable space over a certain averaged measure μ, and transforms it to a generalization problem under the non-iid setting. This might have its own interest in RL theory for better understanding

\epsilon

-greedy exploration in deep RL.

2022