AIRSTREAM: A Neural Network Model of Auditory Scene Analysis and Source Segregation

Multiple sound sources often contain harmonics that overlap and may be degraded by environmental noise. The auditory system is capable of teasing apart these sources into distinct mental objects, or streams. Such an "auditory scene analysis" enables the brain to solve the cocktail party problem. A neural network model of auditory scene analysis, called the AIRSTREAM model, is presented to propose how the brain accomplishes this feat. The model clarifies how the frequency components that correspond to a give acoustic source may be coherently grouped together into distinct streams based on pitch and spatial cues. The model also clarifies how multiple streams may be distinguishes and seperated by the brain. Streams are formed as spectral-pitch resonances that emerge through feedback interactions between frequency-specific spectral representaion of a sound source and its pitch. First, the model transforms a sound into a spatial pattern of frequency-specific activation across a spectral stream layer. The sound has multiple parallel representations at this layer. A sound's spectral representation activates a bottom-up filter that is sensitive to harmonics of the sound's pitch. The filter activates a pitch category which, in turn, activate a top-down expectation that allows one voice or instrument to be tracked through a noisy multiple source environment. Spectral components are suppressed if they do not match harmonics of the top-down expectation that is read-out by the selected pitch, thereby allowing another stream to capture these components, as in the "old-plus-new-heuristic" of Bregman. Multiple simultaneously occuring spectral-pitch resonances can hereby emerge. These resonance and matching mechanisms are specialized versions of Adaptive Resonance Theory, or ART, which clarifies how pitch representations can self-organize durin learning of harmonic bottom-up filters and top-down expectations. The model also clarifies how spatial location cues can help to disambiguate two sources with similar spectral cures. Data are simulated from psychophysical grouping experiments, such as how a tone sweeping upwards in frequency creates a bounce percept by grouping with a downward sweeping tone due to proximity in frequency, even if noise replaces the tones at their interection point. Illusory auditory percepts are also simulated, such as the auditory continuity illusion of a tone continuing through a noise burst even if the tone is not present during the noise, and the scale illusion of Deutsch whereby downward and upward scales presented alternately to the two ears are regrouped based on frequency proximity, leading to a bounce percept. Since related sorts of resonances have been used to quantitatively simulate psychophysical data about speech perception, the model strengthens the hypothesis the ART-like mechanisms are used at multiple levels of the auditory system. Proposals for developing the model to explain more complex streaming data are also provided.

Neuromorphic model for sound source segregation

While humans can easily segregate and track a speaker's voice in a loud noisy environment, most modern speech recognition systems still perform poorly in loud background noise. The computational principles behind auditory source segregation in humans is not yet fully understood. In this dissertation, we develop a computational model for source segregation inspired by auditory processing in the brain. To support the key principles behind the computational model, we conduct a series of electro-encephalography experiments using both simple tone-based stimuli and more natural speech stimulus. Most source segregation algorithms utilize some form of prior information about the target speaker or use more than one simultaneous recording of the noisy speech mixtures. Other methods develop models on the noise characteristics. Source segregation of simultaneous speech mixtures with a single microphone recording and no knowledge of the target speaker is still a challenge. Using the principle of temporal coherence, we develop a novel computational model that exploits the difference in the temporal evolution of features that belong to different sources to perform unsupervised monaural source segregation. While using no prior information about the target speaker, this method can gracefully incorporate knowledge about the target speaker to further enhance the segregation.Through a series of EEG experiments we collect neurological evidence to support the principle behind the model. Aside from its unusual structure and computational innovations, the proposed model provides testable hypotheses of the physiological mechanisms of the remarkable perceptual ability of humans to segregate acoustic sources, and of its psychophysical manifestations in navigating complex sensory environments. Results from EEG experiments provide further insights into the assumptions behind the model and provide motivation for future single unit studies that can provide more direct evidence for the principle of temporal coherence.

Plasticity in the neural coding of auditory space in the mammalian brain

Sound localization relies on the neural processing of monaural and binaural spatial cues that arise from the way sounds interact with the head and external ears. Neurophysiological studies of animals raised with abnormal sensory inputs show that the map of auditory space in the superior colliculus is shaped during development by both auditory and visual experience. An example of this plasticity is provided by monaural occlusion during infancy, which leads to compensatory changes in auditory spatial tuning that tend to preserve the alignment between the neural representations of visual and auditory space. Adaptive changes also take place in sound localization behavior, as demonstrated by the fact that ferrets raised and tested with one ear plugged learn to localize as accurately as control animals. In both cases, these adjustments may involve greater use of monaural spectral cues provided by the other ear. Although plasticity in the auditory space map seems to be restricted to development, adult ferrets show some recovery of sound localization behavior after long-term monaural occlusion. The capacity for behavioral adaptation is, however, task dependent, because auditory spatial acuity and binaural unmasking (a measure of the spatial contribution to the “cocktail party effect”) are permanently impaired by chronically plugging one ear, both in infancy but especially in adulthood. Experience-induced plasticity allows the neural circuitry underlying sound localization to be customized to individual characteristics, such as the size and shape of the head and ears, and to compensate for natural conductive hearing losses, including those associated with middle ear disease in infancy.

Recreational Food Service is Big Business

In the analysis - Recreational Food Service Is Big Business - by Gary Horvath, President, Recreational Foodservice Division, Service America Corporation and Mickey Warner, Associate Professor School of Hospitality Management at Florida International University, Horvath and Warner initially state: “Recreational food service is very different from routine food service management. The authors review the market and the management planning and challenges that create that difference.” Recreational food is loosely defined by the authors as food for special events. These can be one-time events, repeated events that are not on a fixed schedule [i.e. concerts], weekly events such as football-baseball-or basketball games, or other similar venues. Concessions are a large part of these fan based settings. “An anticipated 101,000 fans at a per capita spending of $5-6 [were expected]. A typical concessions menu of hot dogs, popcorn, soda, beer, snacks, novelty foods, candy, and tobacco products comprises this market segment,” say Horvath and Warner in reference to the Super-Bowl XXI football championship game, held in Rose Bowl stadium in Pasadena, California, on January 25, 1987. Some of the article is based upon that event. These food service efforts focus on the individual fan, but do extend to the corporate-organizational level as well. Your authors will have you know that catering is definitely a part of this equation. The monies spent and earned are phenomenal. “Special events of this type attract numerous corporate catering opportunities for companies entertaining VIP guest lists,” the authors inform. “Hospitality tents usually consist of a pregame cocktail party and buffet and a post-game celebration with musical entertainment held in lavishly decorated tents erected at the site. In this case a total of 5,000 covers, at a price of $200 each, for 12-15 separate parties were anticipated.” Horvath and Warner also want you to know that novelties and souvenirs make up an essential part of this, the recreational food service market. “Novelties and souvenirs are a primary market and source of revenue for every stadium food service operator,” say Horvath and Warner. The term, “per capita spending is the measurement used by the industry to evaluate sales potential per attendee at an event,” say the authors. Of course, with the solid revenue figures involved as well as the number of people anticipated for such events, planning is crucial, say Horvath and Warner. Training of staff, purchasing and supply, money and banking, facility access, and equipment, are a few of the elements to be negotiated. Through both graphs and text, Horvath and Warner do provide a fairly detailed outline of what a six-step event plan consists of.

Using electroencephalography signals to control acoustical processing

Current hearing-assistive technology performs poorly in noisy multi-talker conditions. The goal of this thesis was to establish the feasibility of using EEG to guide acoustic processing in such conditions. To attain this goal, this research developed a model via the constructive research method, relying on literature review. Several approaches have revealed improvements in the performance of hearing-assistive devices under multi-talker conditions, namely beamforming spatial filtering, model-based sparse coding shrinkage, and onset enhancement of the speech signal. Prior research has shown that electroencephalography (EEG) signals contain information that concerns whether the person is actively listening, what the listener is listening to, and where the attended sound source is. This thesis constructed a model for using EEG information to control beamforming, model-based sparse coding shrinkage, and onset enhancement of the speech signal. The purpose of this model is to propose a framework for using EEG signals to control sound processing to select a single talker in a noisy environment containing multiple talkers speaking simultaneously. On a theoretical level, the model showed that EEG can control acoustical processing. An analysis of the model identified a requirement for real-time processing and that the model inherits the computationally intensive properties of acoustical processing, although the model itself is low complexity placing a relatively small load on computational resources. A research priority is to develop a prototype that controls hearing-assistive devices with EEG. This thesis concludes highlighting challenges for future research.

Consumer product-related injuries in Queensland children : A snapshot of current problem areas for potential action

This report presents a snapshot from work which was funded by the Queensland Injury Prevention Council in 2010-11 titled “Feasibility of Using Health Data Sources to Inform Product Safety Surveillance in Queensland children”. The project provided an evaluation of the current available evidence-base for identification and surveillance of product-related injuries in children in Queensland and Australia. A comprehensive 300 page report was produced (available at: http://eprints.qut.edu.au/46518/) and a series of recommendations were made which proposed: improvements in the product safety data system, increased utilisation of health data for proactive and reactive surveillance, enhanced collaboration between the health sector and the product safety sector, and improved ability of health data to meet the needs of product safety surveillance. At the conclusion of the project, a Consumer Product Injury Research Advisory group (CPIRAG) was established as a working party to the Queensland Injury Prevention Council (QIPC), to prioritise and advance these recommendations and to work collaboratively with key stakeholders to promote the role of injury data to support product safety policy decisions at the Queensland and national level. This group continues to meet monthly and is comprised of the organisations represented on the second page of this report. One of the key priorities of the CPIRAG group for 2012 was to produce a snapshot report to highlight problem areas for potential action arising out of the larger report. Subsequent funding to write this snapshot report was provided by the Institute for Health and Biomedical Innovation, Injury Prevention and Rehabilitation Domain at QUT in 2012. This work was undertaken by Dr Kirsten McKenzie and researchers from QUT's Centre for Accident Research and Road Safety - Queensland. This snapshot report provides an evidence base for potential further action on a range of children’s products that are significantly represented in injury data. Further information regarding injury hazards, safety advice and regulatory responses are available on the Office of Fair Trading (OFT) Queensland website and the Product Safety Australia websites. Links to these resources are provided for each product reviewed.

On secure multi-party computation in black-box groups

We study the natural problem of secure n-party computation (in the passive, computationally unbounded attack model) of the n-product function f G (x 1,...,x n ) = x 1 ·x 2 ⋯ x n in an arbitrary finite group (G,·), where the input of party P i is x i  ∈ G for i = 1,...,n. For flexibility, we are interested in protocols for f G which require only black-box access to the group G (i.e. the only computations performed by players in the protocol are a group operation, a group inverse, or sampling a uniformly random group element). Our results are as follows. First, on the negative side, we show that if (G,·) is non-abelian and n ≥ 4, then no ⌈n/2⌉-private protocol for computing f G exists. Second, on the positive side, we initiate an approach for construction of black-box protocols for f G based on k-of-k threshold secret sharing schemes, which are efficiently implementable over any black-box group G. We reduce the problem of constructing such protocols to a combinatorial colouring problem in planar graphs. We then give two constructions for such graph colourings. Our first colouring construction gives a protocol with optimal collusion resistance t < n/2, but has exponential communication complexity O(n*2t+1^2/t) group elements (this construction easily extends to general adversary structures). Our second probabilistic colouring construction gives a protocol with (close to optimal) collusion resistance t < n/μ for a graph-related constant μ ≤ 2.948, and has efficient communication complexity O(n*t^2) group elements. Furthermore, we believe that our results can be improved by further study of the associated combinatorial problems.

Tiering Effects in Third-party Logistics: A First-tier Buyer Perspective

This doctoral dissertation takes a buy side perspective to third-party logistics (3PL) providers’ service tiering by applying a linear serial dyadic view to transactions. It takes its point of departure not only from the unalterable focus on the dyad levels as units of analysis and how to manage them, but also the characteristics both creating and determining purposeful conditions for a longer duration. A conceptual framework is proposed and evaluated on its ability to capture logistics service buyers’ perceptions of service tiering. The problem discussed is in the theoretical context of logistics and reflects value appropriation, power dependencies, visibility in linear serial dyads, a movement towards the more market governed modes of transactions (i.e. service tiering) and buyers’ risk perception of broader utilisation of the logistics services market. Service tiering, in a supply chain setting, with the lack of multilateral agreements between supply chain members, is new. The deductive research approach applied, in which theoretically based propositions are empirically tested with quantitative and qualitative data, provides new insight into (contractual) transactions in 3PL. The study findings imply that the understanding of power dependencies and supply chain dynamics in a 3PL context is still in its infancy. The issues found include separation of service responsibilities, supply chain visibility, price-making behaviour and supply chain strategies under changing circumstances or influence of non-immediate supply chain actors. Understanding (or failing to understand) these issues may mean remarkable implications for the industry. Thus, the contingencies may trigger more open-book policies, larger liability scope of 3PL service providers or insourcing of critical logistics activities from the first-tier buyer core business and customer service perspectives. In addition, a sufficient understanding of the issues surrounding service tiering enables proactive responses to devise appropriate supply chain strategies. The author concludes that qualitative research designs, facilitating data collection on multiple supply chain actors, may capture and increase understanding of the impact of broader supply chain strategies. This would enable pattern-matching through an examination of two or more sides of exchange transactions to measure relational symmetries across linear serial dyads. Indeed, the performance of the firm depends not only on how efficiently it cooperates with its partners, but also on how well exchange partners cooperate with an organisation’s own business.

The SDLP and the Europeanization of the Northern Ireland Problem

This article explores the various ways in which the Social Democratic and Labour Party (SDLP) has used Europe – as a source of financial aid, political support, ideas and inspiration – in its attempts to resolve the Northern Ireland conflict. In this, the piece considers the SDLP, not as a subject, but rather as an advocate of the Europeanization of the Northern Ireland problem. In particular, it looks at the role of John Hume, a founding member and later leader of the SDLP, who inculcated a strongly pro-European outlook within the party. In doing so, the article considers the success of Hume and the SDLP in their efforts to bring a European influence to bear on Northern Ireland, especially in relation to the peace process and the 1998 Agreement. However, it also looks at both the limitations of this influence, and the problems involved with the SDLP's pro-European approach, particularly since Hume's departure as party leader in 2001. In conclusion, the article suggests that the party may have been ‘over-Europeanized’, with its long-term focus on European issues and ideas now becoming electorally disadvantageous. In this way, the Europeanization of the Northern Ireland problem, and by extension the SDLP, has proven costly to the party.

Scotland's Sectarianism Problem: Irish Answers?

A series of developments during the 2010–11 football season has led to an intense public debate over the question of the nature and extent of religious sectarianism in Scotland. The Scottish National Party (SNP) government has responded with a new piece of legislation which has been widely criticised and has prompted some commentators to speculate about a political ‘own goal’. This article provides a guide to the debate around sectarianism and its historical and political dimensions. It also suggests that the Irish roots of the problem in Scotland should be properly acknowledged, and that a possible way forward could involve cooperation between Scotland, Northern Ireland and the Republic of Ireland within the structures and procedures of the British–Irish Council (BIC).

WAIS Seminar:Secure Multi-Party Computation for the Masses

Speaker(s): Prof. David Evans Organiser: Dr Tim Chown Time: 22/05/2014 10:45-11:45 Location: B53/4025 Abstract Secure multi-party computation enables two (or more) participants to reliably compute a function that depends on both of their inputs, without revealing those inputs to the other party or needing to trust any other party. It could enable two people who meet at a conference to learn who they known in common without revealing any of their other contacts, or allow a pharmaceutical company to determine the correct dosage of a medication based on a patient’s genome without compromising the privacy of the patient. A general solution to this problem has been known since Yao's pioneering work in the 1980s, but only recently has it become conceivable to use this approach in practice. Over the past few years, my research group has worked towards making secure computation practical for real applications. In this talk, I'll provide a brief introduction to secure computation protocols, describe the techniques we have developed to design scalable and efficient protocols, and share some recent results on improving efficiency and how secure computing applications are developed.