I’m training a neural network to generate recipes based on a database of about 30,000 examples, using the open-source char-rnn framework. People have asked me whether I’ve tried making any of these recipes. The answer is no. No, I most definitely haven’t. I don’t know, in fact, of anyone who has dared
« Previous Page — Table of Contents — Next Page » A major limitation of many methods used for studying large-scale networks stems from the fact that the strength of ties is not taken into account. Granovetter (1973) argued that the strength of a social tie is a function of its duration, emotional intensity, intimacy
Select Page Neural representations of speech How are speech features represented and combined in the brain to form a unified meaning? Our research work aims to uncover the neural correlates of these representations, and the dynamical mechanisms by which they are transformed and integrated. Key publications: Neural manifolds carry reactivation of phonetic representations during semantic processing Interpretability of statistical approaches in speech and language neurosciecne Imagined speech can be decoded fr
Gerardo Duran-Martin cv articles posts # Posts bandits3 Bayes8 bayesian‑deep‑learning1 changepoint1 data‑analysis3 einsums2 ewma1 filtering5 frequentist1 gaussian‑processes1 HMM1 jax1 kalman‑filter3 linear‑algebra1 neural‑networks1 non‑stationary5 regimes3 sequential1 sequential‑decision‑making4 slides2 time‑series1 uncertainty6 ## 2026 A journey through a PhD 13 Apr 2026 Reflections as a PhD student in statistical machine learning. Streaming hidden Markov models 10 Apr 2026 A one
Generative Adversarial Networks (GANs) are a type of simple machine learning algorithmic rule that has gained significant attention in recent years
# hasktorch ## Differentiable Programs (Neural Networks) From a functional programming perspective, a neural network is represented by data and functions, much like any other functional program. The only distinction that differentiates neural networks from any other functional program is that it implements a small interface surface to support differentiation. Thus, we can consider neural networks to be "differentiable functional programming". The data in neural networks are the values to be fitted that p
Neural networks underlie complex brain information processing, yet the role of their single-neuron topology in governing computation and behavior remains unclear, particularly regarding how it shapes individual neuron function and activity evolution during learning. Using two-photon calcium imagi
Convolutional neural networks show promise as models of biological vision. However, their decision behaviour, including the facts that they are deterministic and use equal numbers of computations for easy and difficult stimuli, differs markedly from human decision-making, thus limiting their applicability as models of human perceptual behaviour. Here we develop a new neural network, RTNet, that generates stochastic decisions and human-like response time (RT) distributions. We further performed comprehensive
A collaborative project between human and machine intelligence, exploring new aesthetic possibilities through GANs and pre-trained language models, exhibited at the Water/River art festival in Catalonia.
In this video, we explain the concept of training an artificial neural network