IDSIA: Neural Networks, LSTM and AI Research

The Dalle Molle Institute for Artificial Intelligence in Lugano is jointly associated with USI and SUPSI, with research in neural networks, reinforcement learning, causality and optimization.

kindResearch lab
founded1988

Updated

A joint institute for AI research

Founded in Lugano in 1988 by Angelo Dalle Molle, IDSIA combines AI research and teaching. Its overview describes a joint USI/SUPSI bridge affiliation since 2000, with work spanning neural networks, probabilistic and causal models, robotics, control and optimization. That joint structure does not justify assigning it to one sole parent.

Its neural-network research includes recurrent systems for sequence processing and long-range dependencies. The 1997 LSTM paper by Sepp Hochreiter and Jürgen Schmidhuber describes a gated architecture intended to preserve error signals across long intervals. The bibliographic record identifies Hochreiter's TUM affiliation, so the paper should not be represented as exclusively developed at IDSIA.

LSTM addresses learning dependencies across sequences. IDSIA's present research remit extends beyond that historical paper to reinforcement learning, geometric deep learning and causal reasoning.

Sources

  1. [1]
    Overview - IDSIA

    IDSIA · retrieved Oct 23, 2026

  2. [2]
  3. [3]
    Long short-term memory

    Neural Computation · Nov 15, 1997

IDSIA history

1997

■
LSTM gives networks a memory

With Technische Universität München

Hochreiter and Schmidhuber solved the vanishing gradient problem with a gated cell that could hold information across long sequences. It ran sequence modeling for two decades, until attention took the job.

Neural Computation
Research