AmonI Seminars @amoni-bath - Tumblr Blog

A simple behavioral model that predicts the emergence of complex animal hierarchies

Location: University of Bath, campus, 1W 3.103

Time/date: 11:15-12:15, 27/05/2016

Abstract:

Dominance hierarchies determine how resources are distributed in social animals, including humans. The familiar pecking order ranks animals from most to least dominant, but many groups instead have a single “despot”, while others are dominated by a small oligarchy. How these alternative types arise is still poorly understood. I developed a model that traces each type to a specific combination of three simple interactions: winner-loser contests, policing of contenders by subordinates, and a “winner-winner” interaction that allows some contenders to share dominance. I use the model to explain an example of shared dominance in ants. The model further suggests that simple rules may account for complex social structures, even in more cognitively advanced species. I will also discuss my other research related to collective cognition, in which a group of individuals together processes information and acts as a single cognitive unit.

Speakers’ biography:

Dr. Sasaki is Marie Skłodowska-Curie Postdoctoral Fellow at the University of Oxford and interested in when and how groups make better (or worse) decisions than individuals. His research directly addresses this question, using pigeon homing as a model system. I compare the cognitive abilities of individuals and groups by testing them on the same spatio-cognitive tasks. Thanks to the cutting-edge GPS devices, he can create deatailed agent-based models using high-resolution individual spatiotemporal data. This research is relevant to multiple research fields, including optimal decision-making theory and collective robotics.

Real science online: Engaging 1.5 million volunteers to turn data into discovery

Date/Time: 4:15 pm 31 March 2016

Location: CB 4.8

Speaker: Dr. Ali Swanson, Zooniverse and University of Oxford

As technology changes the way that we collect data, citizen science is changing the way we turn data into discovery. Engaging volunteers to interpret large datasets that computers can’t easily process can expand the scope and scale of scientific research – allowing us to answer bigger, broader questions about the world around us. The Zooniverse citizen science platform engages 1.5 million volunteers around the world in this large-scale data interpretation – on projects that range from plankton to penguins, astronomy to archaeology, cancer research to Martian craters. Using Snapshot Serengeti (www.snapshotserengeti.org) as a case study, I will demonstrate approaches for converting enormous volunteer effort into expert quality scientific data. These approaches have led to >100 peer reviewed publications across disciplines ranging from ecology to climate science to astrophysics. I will describe a newly launched web-based platform that enables all researchers to leverage the power of citizen science to turn data into discovery.

#university of oxford #zooniverse #talk #university of bath #bath #public talk #academia #experiments #data science

The electrical homunculus: ERP studies of somatosensation

Date: 03/02/2016 15:15-16:05

Location: 1W 3.103

Speaker: Dr. Bettina Forster - City University London

Please note, this talk is organised by Dr. Michael J. Proulx, Department of Psychology, University of Bath

Abstract

The somatosensory system not only deals with information coming from our own body but is also crucially involved in embodied cognition. In this talk I will present several recent ERP studies revealing the underlying neural processes when attending to touch on the body and the contribution of somatosensory processes to emotion perception and memory. In particular, I will discuss the neural signature when orienting attention to the body and when selecting a tactile target from distractor events. Further, I will discuss studies showing the effect of body posture on early visual face processing and evidence showing somatotopic encoding of other’s emotion. Finally, I will discuss the contribution of somatosensory cortex to remembering body related information. Taken together, this talk will highlight the diverse aspects of somatosensory processing, discuss its neural signatures and show new approaches to studying the neural basis of embodied cognition.

The Computational Face - Novel approaches in an age of big data

Dates: 03/02/2016: 16:15-17:15 and 04/02/2016 09:15-10:15

Location: University of Bath, main campus: 1WN 3.24

Abstract In these talks, Michel will present recent advances in computer vision and machine learning made by my team at the University of Nottingham.

Wednesday’s talk: Michel will focus in this talk on his work on facial point localisation using direct displacement prediction (regression), and on recent work on using continuous regression to make facial point tracking an order of magnitude faster than the current state of the art.

Thursday’s talk: Michel will focus in this talk on facial expression analysis in general and the potential of a new field of study that combines affective computing, social signal processing, and medicine, which he coined Behaviomedics. Behaviomedics studies medical conditions that cause altered expressive behaviour, which can in itself be objectively measured using computer vision and machine learning techniques, thus contributing to their diagnosis, monitoring, and potential treatment.

Biography Michel Valstar (http://www.cs.nott.ac.uk/~mfv) is an associate professor at the School of Computer Science, University of Nottingham, and member of both the Computer Vision and Mixed Reality Labs. He received his masters' degree in Electrical Engineering at Delft University of Technology in 2005 and his PhD in computer science at Imperial College London in 2008, and he was a Visiting Researcher at MIT's Media Lab. He works in the fields of computer vision and pattern recognition, where his main interest is in automatic recognition of human behaviour, specialising in the analysis of facial expressions. He is the founder of the facial expression recognition challenges (FERA 2011/2015), and the Audio-Visual Emotion recognition Challenge series (AVEC 2011-2015). He is the coordinator of the EU Horizon2020 project ARIA-VALUSPA, which will build the next generation virtual humans, and recipient of Melinda & Bill Gates Foundation funding to help premature babies survive in the developing world. In 2007 he won the BCS British Machine Intelligence Prize for part of his PhD work. His work has received popular press coverage in, among others, Science Magazine, The Guardian, New Scientist and on BBC Radio. He has published over 50 peer-reviewed papers at venues including PAMI, CVPR, ICCV, SMC-Cybernetics, and Transactions on Affective Computing (h-index 28, >3400 citations).

#bigdata #computer vsion #computervision #affection #amoni #talk #seminar

Natural Language Processing: Detection in Deception Spectrum

Date: Monday 30/11/2015 Time: 13:15 - 14:15 Room: 1WN 3.12 Abstract The talk will give a brief introduction to Natural Language Processing (NLP) and Computational Linguistics (CL) and provide few real world scenarios on their applications. Example of such can be seen in relation to Deception and its Detection in various areas - such as online dating and court testimonies - where linguistic and behavioural analysis become the main subject of studies. The talk will then focus on Child Online Grooming and Author Identification - which is concerned with identification of real authors of verbal (more specifically, written) content – and ways in which they might be addressed. Although not new concerns, these fields have been receiving increased attention in recent years due to the Web and Social Media, which allow individuals both to impersonate and to masquerade, effectively anonymously, under multiple digital personas. The talk will conclude by presenting an example of Author Identification where such identification has significant importance.

Biography Dr Anna Vartapetiance, is an Entrepreneur and Researcher at the University of Surrey's Centre for Cyber Security, currently working on Parental Control Systems and Child Online Safety. She was awarded a PhD from the Department of Computer Science at University of Surrey for her work on Deception Detection using Natural Language Processing (NLP) to develop automated/semi-automated systems for detection. Her research has found application in systems that enhance ethical outcomes and issues related to areas such as National DNA databases, Online Gambling, Virtual Worlds and Machine ethics and has been published in over 15 peer reviewed journal papers, proceedings and book chapters. She has received a number of awards for her work including the best speaker prize at the annual London Hopper Colloquium, – an annual conference and networking event for women in computer science, in 2014. Anna is also a Member of International Federation of Information Processing (IFIP) Special Interest Group 9.2.2 Framework on Ethics of Computing (SIG 9.2.2) and Working Group on Social Accountability and computing (WG 9.2) as well as Committee Member of BCS ICT Ethics Specialist Group.

Please, feel free to invite your colleagues.

Costly punishment as a strategy for optimising public goods investment

Date: Monday, 12 October

Time: 12:15 - 13:15

Room: 3E 2.9

Abstract

Contributing to the public good is, perhaps definitionally, a good thing. Why then do some peoplegiven the opportunitypunish those who have contributed to their own (the punisher's) good? The correlation in the expression of this behaviour (known as anti-social punishment) to poor economies and low rule of law makes untangling the causality and mechanisms imperative. Here I present ongoing research into an unusual hypothesis: that the ultimate explanation for costly punishment in humans is both up and down regulation of individual investment at the group level. I will present human data owing primarily to Herrmann, Thvni & Gdchter (2008) with further models and analysis, due primarily to the work of my own group, but rooted in the theoretical biology literature, particularly MacLean, Fuentes-Hernandez, Greig, Hurst & Gudelj (2010). If time allows I will also speculate briefly on implications, specifically that if human sociality has a large implicit component, what impact does and will our changing socio-technological landscape have on human cooperation?

Joanna Bryson is a Reader (tenured Associate Professor) at the University of Bath, currently on sabbatical as a Visiting Fellow at Princeton's Center for Information Technology Policy (CITP). She has broad academic interests in the structure and utility of intelligence, both natural and artificial. She has been publishing on AI ethics since 1998, and been engaged in AI policy in the UK since coauthoring the 2011 EPSRC Principles of Robotics. Her sabbatical project, Public Goods and Artificial Intelligence, includes both basic research in human sociality and experiments in technological interventions. She holds degrees in Psychology from Chicago and Edinburgh, and in Artificial Intelligence from Edinburgh and MIT. At Bath she founded the Intelligent Systems research group (one of four in the Department of Computer Science). She also heads Artificial Models of Natural Intelligence, where she and her colleagues publish in biology, anthropology, cognitive science and systems AI.

Justin Lane (Oxford)

Date: Tuesday 21 July Time 4pm (really!! not 4:15!) Room: 1WN 3.24

Title: Agent Based Simulation of Large-Scale Religious Systems

Abstract: This research utilizes novel methods for agent based modelling in order to simulate the individual and group level dynamics of religious ritual systems. By employing a multi-agent artificial intelligence (Lane, 2013) framework this research modelled the relevant beliefs and behaviours posited by the theory of divergent modes of religiosity (Whitehouse, 2004). It overcomes computational issues by deploying the simulation platform on a high power computing cluster (Oxford’s ARC system). This research both replicates the results of earlier models (McCorkle & Lane, 2012; Whitehouse, Kahn, Hochberg, & Bryson, 2012) and expands these attempts by creating a more generalizable model incorporating both semantic and episodic memory systems as well as social identification mechanisms. This platform allows us to track the social network effects of ritual participation as well as the spread of information within complex representational schemas through both social and ritual interactions.

Miles Brundage

Monday 13th July 2015, 4:15pm, 1W 3.56

Progress in Artificial Intelligence and the Future of Work Many journalists and scholars have recently argued that advances in AI, robotics, and other areas may substantially reduce the demand for many forms of human labor in the economy in the coming decades. A transition to a more labor-light economic system would have wide-ranging social, political, and ethical implications that have only begun to be anticipated, debated, and managed. However, much disagreement still remains about a key premise of such scenarios, namely that technology will increasingly substitute for (and not merely complement) human labor across a wide range of tasks and industries in the coming years, in part thanks to continued progress in AI. In this talk, I'll survey the debate on these questions, focusing in particular on how different estimations of future progress in AI entail different conclusions about the possible futures of work. I'll suggest ways that light could be shed on rates of future AI progress, and how this could inform policy analysis and action on the future of work. Finally, I'll connect issues of automation and work to broader questions of responsibility in AI innovation, suggesting ways that AI researchers might be able to consciously shape the future of work through their research approaches and emphases.

Dr. David Carmel

Thursday 18th June 2015, 1:15pm, 1WN 3.11

Mechanisms of conscious and unconscious perception

The age-old interest in consciousness and subjective experience has evolved in recent years into a major effort to discern the cognitive and neural mechanisms involved in perceptual awareness. What determines the content of our consciousness at any given time? And can meaningful perceptual processing occur for sensory stimuli that do not reach awareness? In this talk I will describe recent work in which I used binocular rivalry, interocular suppression, and attentional manipulations to address these related issues. The first part of the talk will focus on a series of transcranial magnetic stimulation (TMS) studies establishing the importance and specific roles of non-visual, high-level regions of parietal cortex in selecting sensory stimuli for conscious representation. The second part will go into an ongoing series of psychophysiological studies investigating the processing of emotional stimuli in the absence of awareness; this work demonstrates that conscious and unconscious processing differ qualitatively, not only in how they develop over time but also in the specific physiological responses each type of processing evokes. Finally, the third part will present new data on how attentional manipulations can influence perceptual sensitivity at both attended and unattended spatial locations.

Dr. David Carmel is a lecturer at the University of Edinburgh in the school of Philosophy, Psychology, and Language Sciences.

Dominic Mitchell & Paul Rauwolf

Tuesday 16th June 2015, 1:15pm, 1WN 3.24

3 speed talks:

Value Homophily Benefits Cooperation but Motivates Employing Incorrect Social Information - Paul Rauwolf

In Markets, Fairness Evolves Because of Partial Information - Paul Rauwolf

When is Accepting Erroneous Social Information Valuable - Insights into the Winner/Loser Effect - Dominic Mitchell

Value Homophily Benefits Cooperation but Motivates Employing Incorrect Social Information

When an individual discovers in isolation that a socially held view is false, should they alter their view? Empirical evidence shows that humans often do not. Individuals often judge others based on third party gossip, rather than their own experience, despite the fact that gossip is error prone. Rather than judging others on their merits, even when such knowledge is free, we judge based on the opinions of third parties. The null hypothesis is that such behaviours are maladaptive exceptions to what is typically an adaptive heuristic. Social learning could be the best strategy despite a high incidence of error when the full cost of accruing accurate information, including time, is taken into account. Our research seeks to understand whether, in some social contexts, it is actually advantageous to use incorrect social information in lieu of the truth.

In Markets, Fairness Evolves Because of Partial Information

The trust game offers an individual the opportunity to invest with another, in hopes the other can be trusted to return the investment. When offered the opportunity to invest, humans tend to make two, seemingly paradoxical, errors. First, when an investor possesses no information about whether the trade will be beneficial, they trust and invest anyway. Second, when they know the trade will be beneficial, they reject it, unless the trade is fair. Why would it be useful to reject advantageous offers, but accept offers of unknown quality? Here, we unify these two behaviors. We find that in a market place with partial information it is advantageous to both: 1.) trust unknown partners, and 2.) demand fairness from known partners.

When is Accepting Erroneous Social Information Valuable - Insights into the Winner/Loser Effect

(abstract forthcoming)

Joint IS/HCI Affective Computing Workshop

Tuesday 19th May 2015, 1:15pm, 3W 3.7

Workshop held in Ignite Format, with each speaker providing a short, 3 minute presentation, followed by an open discussion session.

Speakers

Dave Brown (Psychology) Joanna Bryson (Computer Science) Nataliya Mogles (Computer Science ) Julian Padget (Computer Science) Karin Petrini (Psychology) Matthew Thompson (Computer Science) Leon Watts (Computer Science)

Organisers

Denise Lengyel, Rob Wortham

Yifei Wang

Thursday 30th April 2015, 2:15pm, 4W 1.7

Recombination Is Surprisingly Constructive for Artificial Gene Regulatory Networks in the Context of Selection for Developmental Stability

Recombination is ubiquitous in multicellular plants, animals and even fungus. Many studies have shown that recombination can generate a great amount of genetic innovations, but it is also believed to damage well-adapted lineages, causing debates over how organisms cope with such disruptions.

Using an established model of artificial gene regulatory networks, here we show that recombination may not be as destructive as expected. Provided only that there is selection for developmental stability, recombination can establish and maintain lineages with reliably better phenotypes compared to asexual reproduction. Contrary to expectation, this does not appear to be a simple side effect of higher levels of variation.

A simple model of the underlying dynamics demonstrates a surprisingly high robustness in these lineages against the disruption caused by recombination. Contrary to expectation, lineages subject to recombination are less likely to produce offspring suffering truncation selection for instability than asexual lineages subject to simple mutation. These findings indicate the fundamental differences between recombination and high mutation rates, which has important implications for understanding both biological innovation and hierarchically structured models of machine learning.

Yifei Wang, PhD Student, Department of Computer Science, University of Bath

Dr Patrick Crogan

Thursday 30th April 2015, 1W 2.103, 1:15pm

AI, Technicity and Souls: Military AI and Automation

This talk will look at AI from a broad perspective as

1. Technological development (philosophy of technology), and

2. A legacy of 20th century military technoscientific invention from the fertile post-1945 period (critical media/cultural history).

The goal will be to lay out some lines of approach to AI at a time when automatic and automated systems are increasingly populating the contemporary and future landscape. The focus is on recent efforts to promote and develop automated lethal autonomous robots.

Dr Patrick Crogan, Associate Professor of Digital Cultures from the Digital Cultures Research Centre, University of the West of England

Paul Rauwolf

Thursday 19th March 2015, 1W 2.104, 1:15pm

The Evolution of the Impact Bias

There is robust evidence that prior to an anticipated event, people systematically fail to accurately predict their feeling after the event. They predict more intense emotions compared to their actual affective experiences, a phenomena known as impact bias. Furthermore, individuals do not improve their predictions by learning, despite previous experience. Many explanations for impact bias have been hypothesized, including biased memories, overemphasis of one’s current focus (focalism bias), and limitations in imagining future events. These account for the bias’ mechanism (proximate explanation), but not how the bias survives natural selection (ultimate explanation). We hypothesize an ultimate explanation: impact bias helps humans navigate decision making in noisy environments.

Paul Rauwolf, PhD Student, Dept of Computer Science, University of Bath

Dr. Richard Watson

Thursday 12th March 2015, 1W 2.103, 1:15pm

Understanding the evolution of evolvability using the theory of memory, learning and generalisation.

In order for evolution by natural selection to explain the adaptation that we observe in natural populations we must account for the availability of suitable variations that natural selection can act on. Rupert Riedl, an early pioneer of evolutionary developmental biology, suggested that this is facilitated by a specific developmental organisation that is itself a product of past selection. However, the construction of a theoretical framework to formalise such ‘evolution of evolvability’ has been continually frustrated by the indisputable fact that natural selection cannot favour structures for benefits they have not yet produced. Here we resolve this problem. Recent work shows that short-term selective pressures on gene interactions are functionally equivalent to a simple type of associative learning, well-understood in neural network research. This is important for the evolution of evolvability because this type of learning system can clearly change in a way that improves its performance on future test cases, i.e. before it has been exposed to those cases, without the need for the future to cause the past. Recognising a formal link with the conditions that enable such predictive generalisation in learning systems unlocks well-established theory to apply to understanding the evolution of evolvability. Here we use this to elucidate, and demonstrate for the first time, conditions where short-term selective pressures (producing the organisation that Riedl suggested) alter evolutionary trajectories in a manner that systematically improves long-term evolutionary outcomes. More generally, we discuss and demonstrate how learning theory can be converted to the evolutionary domain to demystify theoretical problems in evo-devo, evo-eco and the major evolutionary transitions.

Dr. Richard A. Watson, Associate Professor, Institute for Life Sciences/Electronics and Computer Science, University of Southampton

Dr Kathleen Richardson

Thursday 5th March 2015, 1WN 3.24, 1:15pm

Making Others: The Human in the Ethics of Robotics

This talk will investigate the anthropology of making humans and robots and bring these two ways of making together to explore what it means to be human. Are humans carrying out more than a mechanical exchange of services when they interact with each other? As humanoid robots are socially designed platforms inspired by existing human relational practices, what is the traffic that flows from human to robot and back to the human?

Robotic scientists imagine robots in the role of the other. If humans make humans (as children and adults) and technologists are making robots (mechanical humans), what kinds of humans are made as a result of robotic interactions? Are robotic machines just mimicking the roles the humans already play in relational encounters?

The talk will explore those areas where robots are imagined to act as alternatives to humans, in the service industry, sex-work and for human attachment difficulties.

Dr Kathleen Richardson, Senior Research Fellow in the Ethics of Robotics, Centre for Computing and Social Responsibility, De Montfort University

Dr Ron Chrisley

Thursday 19th February 2015, 1W 2.103, 1:15pm

Beyond Predictive Coding: An Expectation-based Account of Consciousness

It has been recently suggested that predictive coding models, originally developed to account for general cognitive and perceptual phenomena, might also be used to explain various aspects of consciousness. I argue that despite the appeal of the notion of expectation central to predictive coding models, such models typically also involve a number of further claims that unnecessarily constrain, and perhaps disadvantage, expectation-based explanations of consciousness. For example: 1) the benefits of expectation-based accounts of consciousness can be de-coupled from hypotheses concerning (e.g. Bayesian) optimality; 2) predictive coding models are typically not predictive in any direct sense, in that they merely involve representations of the causes of sensory input, thus forfeiting a distinctive aspect of expectation-based approaches to experiential content; 3) predictive coding models often fail to involve action in an essential way, accruing the familiar limitations of disembodied accounts of mind; and 4) the predictive coding approach of a "winner-take-all" promotion of a single model to be the content of consciousness prevents it from accommodating anti-realist or at least more critically realist views of consciousness (cf, e.g., Dennett 1991). I argue that a more general expectation-based model of consciousness can preserve the strengths of predictive coding models, while avoiding the above limitations.

Dr Ron Chrisley, Reader in Philosophy, Sackler Centre for Consciousness Science, University of Sussex.

Trending Blogs

Recently Viewed Blogs

AmonI Seminars