Please upgrade your browser

For the best experience, you should upgrade your browser. Visit our accessibility page to view a list of supported browsers along with links to download the latest version.

Student Showcase Archive

Malgorzata Starzynska

MPhil work

MPhil work

Architecture Of Machine Dreams: Convolutional Neural Networks For Point Cloud Shape Recognition

Ever since Brunelleschi used mathematical perspective to translate space into its subdimensional representation, the western ocular experience has been expanding upon this visual concept. But what seemed to have been a consistent development of the dominant Florentian paradigm has now come to a halt. In today’s digital era modes of visual perception are shifting and subsequently changing the way reality is observed. In the words of the Italian philosopher Matteo Pasquinelli, twenty-first-century visual perception no longer belongs to the observer. Attention is replaced by a selective filter, vision by pattern recognition and anomaly detection. In today’s algorithmic reality, perception is no longer an anthropocentric domain.

The current abundance of Big Data will make machine learning an inevitable part of the design process, however, this new algorithmic vision offers more than just an opportunity for generative design. It opens up an opportunity for a fully integrated methodology where artificial neural networks, if appreciated for their perceptual capability, can contribute to the critical analysis of contemporary culture.

The quantity of data has resulted in the unavoidable: a crisis of representation. It will be impossible to address this problem until the need for “new epistemic eye” has been resolved. The proposal aspires to develop an alternative cognitive perspective for this new digital landscape. Informed by works of theoreticians and philosophers concerned with cybernetics and technosphere as well as the theory of semiotics, this project focuses on a tangible proposition; spaces generated by a self-taught AI.

Machine learning will be considered as a tool as well as a vehicle for the critique. A trained algorithm will classify digital models of architectural spaces by looking for pattern commonalities. After classification, the network will be “reversed” to produce generative designs that reflect the learning mechanism behind classification. This process will be based around the premise of second-order cybernetics, where the observer is observing a system in which he himself is participating. In this process, architectural space will become conceptualised into its pure, mathematical form; an abstraction of data. This digital experience of reality will then be examined in an effort to identify the differences between the traditional reading of architectural form and emerging patterns identified by a cognitive algorithm.



  • Malgorzata Starzynska
  • MPhil


    School of Architecture


    Architecture Research, 2018–

  • Malgorzata Starzynska is an architect, an artist, and an educator. Trained in Poland and the UK, she has practiced in London, Zurich, and Beijing. Malgorzata has worked as a guest tutor at the University of Greenwich where she also continues to assist with BA and MArch crits since 2014.

    Malgorzata graduated from BA Architecture with First Class Honours from London South Bank University in 2011 and gained her MArch from the University of Greenwich in 2014. During her BA she was awarded a Student Prize for Recognition of Excellence by the RIBA South London Society of Architecture. Her MArch thesis dealt with the visualisation of scientific concepts, particularly using the imagery of Large Hadron Collider and her final project based on critical findings of her thesis was exhibited at the 72nd World Science Fiction convention.

    Majority of Malgorzata's creative and theoretical work is an expression of her prolonged interest in scientific methodology and its application in architectural theory and practice. In her current research, Malgorzata explores the transdisciplinary advancement in AI and applications of machine learning in 3D space recognition.

  • Degrees

  • Postgraduate Diploma in Architectural Practice, RIBA pt. III, University of Greenwich, 2017; Master of Architecture, RIBA pt.II, University of Greenwich, 2014; Bachelor of Arts in Architecture, RIBA pt. I, London South Bank University, 2011
  • Exhibitions

  • ‘The Gold Mine’, 72nd World Science Fiction convention, LonCon, ExCel Exhibition Centre, London, 2014
  • Awards

  • Student Prize for Recognition of Excellence, South London Society of Architecture, RIBA, 2009