Fernando Espinosa Iñiguez

Hi I'll be at #NeurIPS2025 in San Diego this coming week, excited to meet new friends and learn new things :) #AudioML

Also going to supernova, the time difference from the west coast even is crazy

CoShNet: A Hybrid Complex Valued Neural Network using Shearlets In a hybrid neural network, the expensive convolutional layers are replaced by a non-trainable fixed transform with a great reduction in parameters. In previous works, good results were obtained by re...

arxiv.org/abs/2208.06882

Vortices, topology and time We relate physical time with the topology of magnetic field vortices. We base ourselves on a formulation of unimodular gravity where the cosmological …

www.sciencedirect.com/science/arti...

Directed Semi-Simplicial Learning with Applications to Brain Activity Decoding Graph Neural Networks (GNNs) excel at learning from pairwise interactions but often overlook multi-way and hierarchical relationships. Topological Deep Learning (TDL) addresses this limitation by leve...

Directed Semi-Simplicial Learning with Applications to Brain Activity Decoding arxiv.org/abs/2505.17939

A Geometric Perspective on Variational Autoencoders
proceedings.neurips.cc/paper_files/...

How does Antarctic ice deform? A deep-learning model infers large-scale dynamics of Antarctic ice shelves

In Science, researchers report a physics-informed #DeepLearning model that can predict the deformation behavior of Antarctic ice shelves, revealing complexities of the process that extend beyond the traditional understanding.

Learn more in a new #SciencePerspective. scim.ag/3RgudLc

Time-resolved aperiodic and oscillatory dynamics during human visual memory encoding Biological neural networks translate sensory information into neural code that is held in memory over long timescales. Theories for how this occurs often posit a functional role of neural oscillations...

www.jneurosci.org/content/earl...

Hilbert's sixth problem: derivation of fluid equations via Boltzmann's kinetic theory In this paper, we rigorously derive the fundamental PDEs of fluid mechanics, such as the compressible Euler and incompressible Navier-Stokes-Fourier equations, starting from the hard sphere particle s...

arxiv.org/abs/2503.01800

The Many Faces of Information Geometry (2022)
www.ams.org/notices/2022...

Manifold Topological Deep Learning for Biomedical Data Recently, topological deep learning (TDL), which integrates algebraic topology with deep neural networks, has achieved tremendous success in processing point-cloud data, emerging as a promising paradi...

arxiv.org/abs/2503.00175

Gravity from entropy Gravity is derived from an entropic action coupling matter fields with geometry. The fundamental idea is to relate the metric of Lorentzian spacetime to a quantum operator, playing the role of an reno...

arxiv.org/abs/2408.14391

Can one hear the shape of a drum? (1966)

Symmetries of Living Systems: Symmetry Fibrations and Synchronization in Biological Networks A symmetry is a `change without change'. As simple as it sounds, this concept is the fundamental cornerstone that unifies all branches of theoretical physics. Virtually all physical laws -- ranging fr...

arxiv.org/abs/2502.18713

Sheaf theory: from deep geometry to deep learning This paper provides an overview of the applications of sheaf theory in deep learning, data science, and computer science in general. The primary text of this work serves as a friendly introduction to ...

arxiv.org/abs/2502.15476

Graph Curvature Flow-Based Masked Attention Graph neural networks (GNNs) have revolutionized drug discovery in chemistry and biology, enhancing efficiency and reducing resource demands. However, classical GNNs often struggle to capture long-ran...

pubs.acs.org/doi/pdf/10.1...

A Dynamical Systems View of Psychiatric Disorders—Theory This narrative review describes a new approach to the diagnosis and treatment of psychiatric disorders that is based on dynamical systems theory, which addresses the concepts of tipping points, cycles...

jamanetwork.com/journals/jam...

Dirac-Equation Signal Processing: Physics Boosts Topological Machine Learning Topological signals are variables or features associated with both nodes and edges of a network. Recently, in the context of Topological Machine Learning, great attention has been devoted to signal pr...

arxiv.org/abs/2412.05132

Fourier Features Let Networks Learn High Frequency Functions in Low Dimensional Domains We show that passing input points through a simple Fourier feature mapping enables a multilayer perceptron (MLP) to learn high-frequency functions in low-dimensional problem domains. These results she...

arxiv.org/abs/2006.10739

The Thousand Brains Project: A New Paradigm for Sensorimotor Intelligence Artificial intelligence has advanced rapidly in the last decade, driven primarily by progress in the scale of deep-learning systems. Despite these advances, the creation of intelligent systems that ca...

I follow Jeff Hawkins and the numenta group's work closely and recommend anyone into Neuroscience check it out
arxiv.org/abs/2412.18354

Principled neuromorphic reservoir computing - Nature Communications Reservoir computing designs recurrent networks that simultaneously buffer inputs and form nonlinear features. Here, authors propose a configurable scheme with better scaling where memory buffer and no...

Reservoir 🤝 Neuromorphic
www.nature.com/articles/s41...

A Hodge-FAST Framework for High-Resolution Dynamic Functional Connectivity Analysis of Higher Order Interactions in EEG Signals We introduce a novel framework that integrates Hodge decomposition with Filtered Average Short-Term (FAST) functional connectivity to analyze dynamic functional connectivity (DFC) in EEG signals. This...

I think a lot of the tda techniques being applied to eeg/fmri are especially relevant to future audio research because in practice those signals are noisy 👀
arxiv.org/abs/2502.00249

On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models This paper addresses the general problem of reinforcement learning (RL) in partially observable environments. In 2013, our large RL recurrent neural networks (RNNs) learned from scratch to drive simul...

Re: deepseek
arxiv.org/abs/1511.09249

proceedings.mlr.press/v196/fitz22a...

Topology shapes dynamics of higher-order networks - Nature Physics Higher-order interactions reveal new aspects of the interplay between topology and dynamics in complex systems. This Perspective describes the emerging field of higher-order topological dynamics and d...

🍩
www.nature.com/articles/s41...

Deep-learning enabled generalized inverse design of multi-port radio-frequency and sub-terahertz passives and integrated circuits - Nature Communications Authors report an AI enabled design method for the synthesis of radio-frequency and millimetre-wave integrated circuits. It can discover architectures beyond human intuition and synthesizes these circ...

Just gonna start posting cool papers on here, with or without commentary :) www.nature.com/articles/s41...

Every skill has local minima you get caught in early on. You hit a wall where more practice doesn't mean you get better. To get out of it, you either take steps in random directions and hope you get out of it eventually, or someone else (mentor, peer) tells you what you could be doing better.

Yess we need more of us!

I'd love to be included in this or a future starter pack

would love to be included in this, I'm working in speech & singing ML