Continuous-time recurrent neural networks can robustly memorize and autonomously reproduce arbitrary random spike train patterns with stable accurate relative timing of all spikes, within some range of parameters.
This paper proposes novel global and regional stability analysis conditions based on linear matrix inequalities for a general class of recurrent neural networks. These conditions can be used for state-feedback control design.
명시적 NeRF 모델의 압축 연구를 위해 다양한 기하학, 텍스처 및 재질 복잡성을 가진 3D 객체를 사용하여 Explicit-NeRF-QA 데이터셋을 구축하였다. 이를 통해 NeRF 모델의 고유한 왜곡 유형을 체계적으로 정의하고, 기존 객관적 화질 평가 지표의 성능을 평가하였다.
The proposed pruned autoencoder (pAE) model effectively simulates the lateral geniculate nucleus (LGN) function by integrating feedforward and feedback streams from/to the primary visual cortex (V1), outperforming other models and human benchmarks in visual object categorization tasks.
A novel NeRF-based method that simultaneously combines lip-syncing to a target audio with facial expression transfer to generate high-fidelity talking faces.
This paper proposes Hamiltonian Learning, a novel unified framework for learning with neural networks from a possibly infinite stream of data, in an online manner, without having access to future information.
This study introduces an artificial neural network (ANN) for image classification that is inspired by the aversive olfactory learning circuits of the nematode Caenorhabditis elegans, demonstrating superior performance compared to control networks.
To achieve highly synchronized and realistic speech-driven talking head synthesis, SyncTalk effectively maintains subject identity, enhances synchronization of lip movements, facial expressions, and head poses, and improves visual quality through a novel NeRF-based framework.
単一の動的ニューラルユニットが、同一の動的軌道の中で、時間によって異なる非線形計算を実行できる。
A novel graph-based performance predictor that leverages both forward and reverse representations of neural architectures to enhance prediction accuracy, especially in data-limited settings.