BEGIN:VCALENDAR
VERSION:2.0
PRODID:Icfo
X-PUBLISHED-TTL:P1W
BEGIN:VEVENT
UID:69f27c1bcf4aa
DTSTART:20210325T100000Z
SEQUENCE:0
TRANSP:OPAQUE
LOCATION:ICFO Auditorium and Online (Teams)
SUMMARY:ICFO | PATRICK HÜMBELI
CLASS:PUBLIC
DESCRIPTION:Research at the intersection of machine learning (ML) and quan-
  tum physics is a recent growing field due to the enormous expec- tations 
 and the success of both fields. ML is arguably one of the most promising t
 echnologies that has and will continue to dis- rupt many aspects of our li
 ves. The way we do research is almost certainly no exception and ML\, with
  its unprecedented ability to find hidden patterns in data\, will be assis
 ting future scientific discoveries. Quantum physics on the other side\, ev
 en though it is sometimes not entirely intuitive\, is one of the most succ
 essful physical theories and we are on the verge of adopting some quan- tu
 m technologies in our daily life. Quantum many-body physics is a subfield 
 of quantum physics where we study the collective behavior of particles or 
 atoms and the emergence of phenomena that are due to this collective behav
 ior\, such as phases of matter. The study of phase transitions of these sy
 stems often requires some intuition of how we can quantify the order param
 eter of a phase. ML algorithms can imitate something similar to intu- itio
 n by inferring knowledge from example data. They can\, there- fore\, disco
 ver patterns that are invisible to the human eye which makes them excellen
 t candidates to study phase transitions. At the same time\, quantum device
 s are known to be able to perform some computational task exponentially fa
 ster than classical com- puters and they are able to produce data patterns
  that are hard to simulate on classical computers. Therefore\, there is th
 e hope that ML algorithms run on quantum devices show an advantage over th
 eir classical analog.\nThis thesis is devoted to study two different paths
  along the front lines of ML and quantum physics. On one side we study the
  use of neural networks (NN) to classify phases of mater in many-body quan
 tum systems. On the other side\, we study ML algorithms that run on quantu
 m computers. The connection be- tween ML for quantum physics and quantum p
 hysics for ML in this thesis is an emerging subfield in ML\, the interpret
 ability of learning algorithms. A crucial ingredient in the study of phase
  transitions with NNs is a better understanding of the predictions of the 
 NN\, to eventually infer a model of the quantum system and interpretabilit
 y can assist us in this endeavor. The interpretabil- ity method that we st
 udy analyzes the influence of the training points on a test prediction and
  it depends on the curvature of the NN loss landscape. This further inspir
 ed an in-depth study of the loss of quantum machine learning (QML) applica
 tions which we as well will discuss.\nIn this thesis we give answers to th
 e questions of how we can leverage NNs to classify phases of matter and we
  use a method that allows to do domain adaptation to transfer the learned 
 \"in- tuition\" from systems without noise onto systems with noise. To map
  the phase diagram of quantum many-body systems in a fully unsupervised ma
 nner\, we study a method known from anomaly detection that allows us to re
 duce the human input to a mini- mum. We will as well use interpretability 
 methods to study NNs that are trained to distinguish phases of matter to u
 nderstand if the NNs are learning something similar to an order parame- te
 r and if their way of learning can be made more accessible to humans. And 
 finally\, inspired by the interpretability of classical NNs\, we develop t
 ools to study the loss landscapes of variational quantum circuits to ident
 ify possible differences between classi- cal and quantum ML algorithms tha
 t might be leveraged for a quantum advantage.
DTSTAMP:20260429T214603Z
END:VEVENT
END:VCALENDAR