• Explain the broad concepts behind deep learning from both computer science and neuroscience perspectives. (Assessed in labs and exams).
• Explain deep learning’s advantages and limitations compared to other modelling and machine learning approaches. (Assessed in exams).
• Identify problems that deep learning is suited to addressing in the fields of cognitive (neuro-) science and artificial intelligence. (Assessed in labs and exams).
• Design and implement deep learning approaches to address some problems in the domain of image processing. (Assessed in labs).
• Understand the key concepts and measures of social network and the models of network formation. (Assessed in labs and exams).
• Explain the factors of social influences and diffusions, their differences and complements. (Assessed in exams).
• Relate the social contagion mechanisms to understand the social drivers of real-world problems. (Assessed in labs and exams).
• Perform social network analysis to investigate social structures. (Assessed in labs).

1. Students’ understanding of lectures and reading assignments will be assessed in a final exam that determines 40% of the final grade.
2. Individual assignments will be graded for depth and completion, with each of the two assignments determining 10% of the final grade
3. Group lab assignments will be graded for depth and completion, with each of the two assignments determining 20% of the final grade.
4. You are required to average a passing grade (5.5) across the exam and individual assignments to pass the course. 

Students will attend eight lectures.

The first four lectures will focus on image processing and the human visual system, viewing the visual system as a deep network. We will first introduce the basic processing mechanisms of computer deep learning systems. We will then compare this to the mechanisms of neural processing implemented in biological brains. We will then introduce the main applications of deep learning to visual cognitive science, largely as a model of biological neural systems. Finally, we will see how recent advances in artificial deep learning system more closely model biological neural processing, improving simulations of biological neural systems and giving deep networks new abilities and applications.

In the second part of the course, we will introduce important concepts and challenges in social network analysis and modelling. We will first go through the basic concepts of social network and its measures such as centrality, core-ness, clustering and path length. We will then study the theories and models to explain the formation of social network. Next we will study the contagion process within a network. Focus will be on the simple and complex contagion theories and their explanation in the spread of disease and behaviour. Lastly we will see how we can either minimize or maximize the diffusion process based upon the advances in diffusion models and influence prediction.

Furthermore, students will work through two lab practical assignments, one on visual processing and one on social network analysis. Students will work in groups of 4 in these assignments, with teachers supervising and grading their progress.
Finally, each student will complete a short individual assignment related to each lab assignment.
All parts of the course will be supported by reading assignments.

The lectures will cover the following topics:

• Lecture 1: Principles of deep learning in artificial networks
• Lecture 2: Deep learning in biological neurons and networks
• Lecture 3: Early and feedforward visual processing
• Lecture 4: Higher and recurrent visual processing
• Lecture 5: Social network and its measures
• Lecture 6: Network formation
• Lecture 7: Simple and complex contagion
• Lecture 8: Influence manipulation

Course form
Lectures, group work, assignments.

Literature
Reading assignments (in recommended order):

Part 1:

• Kay KN, Naselaris T, Prenger RJ, Gallant JL (2008) "Identifying natural images from human brain activity". Nature, 452 (7185): 352-355.
• Huth AG, Nishimoto S, Vu AT, Gallant JL (2012) "A continuous semantic space describes the representation of thousands of object and action categories across the human brain". Neuron, 76(6):1210-24.
• Yamins, D. L., H. Hong, C. F. Cadieu, E. A. Solomon, D. Seibert and J. J. DiCarlo (2014). "Performance-optimized hierarchical models predict neural responses in higher visual cortex." Proc Natl Acad Sci U S A 111(23): 8619-8624.
• The above paper is very important, but some students may find it very difficult. If so, read this paper first: Yamins DL, DiCarlo JJ (2016) "Using goal-driven deep learning models to understand sensory cortex". Nature Neuroscience,19(3): 356-65.
• Huth AG, de Heer WA, Griffiths TL, Theunissen FE, Gallant JL (2016) Natural speech reveals the semantic maps that tile human cerebral cortex. Nature. 532(7600):453-8.

Part 2:

•  Watts, D. J.; Strogatz, S. H. (1998) "Collective dynamics of 'small-world' networks" Nature. 393 (6684): 440–442.
• Centola, D. (2010). “The Spread of Behavior in an Online Social Network Experiment”. Science 329 (5996): 1194-1197.
•  Watts, D.J. (2002). “A simple model of global cascades on random networks”. Proc Natl Acad Sci U S A 99 (9): 5766-5771.
• Lacopini, L., Petri, G., Barrat, A., Latora, V. (2019). “Simplicial models of social contagion”. Nature Communications, 10:2485.
• Kitsak, M., Gallos, L.K., Havlin, S., Liljeros, F., Muchnik, L., Stanley, H.E., Makse H.A.(2010). “Identificaiton of influential spreaders in complex networks”. Nature Physics, 6: 888-893.