Toggle light / dark theme

Dr. Sanjeev Namjoshi, a machine learning engineer who recently submitted a book on Active Inference to MIT Press, discusses the theoretical foundations and practical applications of Active Inference, the Free Energy Principle (FEP), and Bayesian mechanics. He explains how these frameworks describe how biological and artificial systems maintain stability by minimizing uncertainty about their environment.

Namjoshi traces the evolution of these fields from early 2000s neuroscience research to current developments, highlighting how Active Inference provides a unified framework for perception and action through variational free energy minimization. He contrasts this with traditional machine learning approaches, emphasizing Active Inference’s natural capacity for exploration and curiosity through epistemic value.

The discussion covers key technical concepts like Markov blankets.
generative models, and the distinction between continuous and discrete implementations. Namjoshi explains how Active Inference moved from continuous state-space models (2003−2013) to discrete formulations (2015-present) to better handle planning problems.

He sees Active Inference as being at a similar stage to deep learning in the early 2000s — poised for significant breakthroughs but requiring better tools and wider adoption. While acknowledging current computational challenges, he emphasizes Active Inference’s potential advantages over reinforcement learning, particularly its principled approach to exploration and planning.

Namjoshi advocates for balanced oversight that enables innovation while maintaining appropriate safeguards. He expresses particular concern about the rapid pace of AI development potentially outpacing our understanding of risks and regulatory frameworks.

Dr. Sanjeev Namjoshi.

Researchers explore an intriguing phenomenon in quantum systems, drawing inspiration from a recent quantum computing experiment.


Earlier this year, researchers at the Flatiron Institute’s Center for Computational Quantum Physics (CCQ) announced that they had successfully used a classical computer and sophisticated mathematical models to thoroughly outperform a quantum computer on a task that some thought only quantum computers could solve.

In a particle collider at CERN, a rarely-seen event is bringing us tantalizingly close to the brink of new physics.

From years of running what is known as the NA62 experiment, particle physicist Cristina Lazzeroni of the University of Birmingham in the UK and her colleagues have now established, experimentally observed, and measured the decay of a charged kaon particle into a charged pion and a neutrino-antineutrino pair. The researchers have presented their findings at a CERN seminar.

It’s exciting stuff. The reason the team has been pursuing this very specific kind of decay channel so relentlessly for more than a decade is because it’s what is known as a “golden” channel, meaning not only is it incredibly rare, but also well predicted by the complex mathematics making up the Standard Model of physics.

Based on a new mathematical framework and large multi-year multi-mission data sets, we reconstruct electric currents and magnetic fields around the dayside magnetopause and their dependence on the incoming solar wind, IMF, and geodipole tilt. The model architecture builds on previously developed mathematical frameworks and includes two separate blocks: for the magnetosheath and for the adjacent outer magnetosphere. Accordingly, the model is developed in two stages: 1) reconstruction of a best-fit magnetopause and underlying dayside magnetosphere, based on a simple shielded configuration, and 2) derivation of the magnetosheath magnetic field, represented by a sum of toroidal and poloidal terms, each expanded into spherical harmonic series of angular coordinates and powers of normal distance from the boundary. The spacecraft database covers the period from 1995 through 2022 and is composed of data from Geotail, Cluster, Themis, and MMS, with the total number of 1-min averages about 3 M. The modeling reveals orderly patterns of the IMF draping around the magnetosphere and of the magnetopause currents, controlled by the IMF orientation, solar wind pressure, and the Earth’s dipole tilt. The obtained results are discussed in terms of the magnetosheath flux pile-up and the dayside magnetosphere erosion during periods of northward or southward IMF, respectively.

The dayside magnetosheath and magnetopause play a principal role in the magnetosphere response to the interplanetary plasma flow. They serve as a main gateway where the first contact occurs between the incoming magnetized solar wind and the geomagnetic field, eventually resulting in a complex chain of magnetospheric processes. Of primary importance here is the mutual orientation of the external IMF and the internal magnetospheric field, defining the reconnection pattern at the boundary. This subject has long been at the center of many studies and extensive debates in the literature, starting from the seminal ideas of Dungey (Dungey, 1961) and followed by a multitude of works, recently summarized in reviews (Trattner et al., 2021; Fuselier et al., 2024). The reconnection geometry has been traditionally addressed in the framework of two basic concepts: the component and antiparallel merging (e.g. (Fuselier et al., 2021), and refs. therein (Qudsi et al., 2023)).