Bioinformatics is rapidly advancing through the "post-genomic" era following the sequencing of the human genome. In preparation for studying the inner workings behind genes, proteins and even smaller biological elements, several subdivisions of bioinformatics have developed. The subdivision of proteomics, concerning the structure and function of proteins, has been aided by the mass spectrometry data source. Biofluid or tissue samples are rapidly assayed for their protein composition. The resulting mass spectra are analyzed using machine learning techniques to discover reliable patterns which discriminate samples from two populations, for example, healthy or diseased, or treatment responders versus non-responders. However, this data source is imperfect and faces several challenges: unwanted variability arising from the data collection process, obtaining a robust discriminative model that generalizes well to future data, and validating a predictive pattern statistically and biologically. This thesis presents several techniques which attempt to intelligently deal with the problems facing each stage of the analytical process. First, an automatic preprocessing method selection system is demonstrated. This system learns from data and selects a combination of preprocessing methods which is most appropriate for the task at hand. This reduces the noise affecting potential predictive patterns. Our results suggest that this method can help adapt to data from different technologies, improving downstream predictive performance. Next, the issues of feature selection and predictive modeling are revisited with respect to the unique challenges posed by proteomic profile data. Approaches to model selection through kernel learning are also investigated. Key insights are obtained for designing the feature selection and predictive modeling portion of the analytical framework. Finally, methods for interpreting the results of predictive modeling are demonstrated. These methods are used to assure the user of various desirable properties: validation of the strength of a predictive model, validation of reproducible signal across multiple data generation sessions and generalizability of predictive models to future data. A method for labeling profile features with biological identities is also presented, which aids in the interpretation of the data. Overall, these novel techniques give the protein profiling community additional support and leverage to aid the predictive capability of the technology.

Students who are solving open-ended problems would benefit from formative assessment, i.e., from receiving helpful feedback and from having an instructor who is informed about their level of performance. Open-ended problems challenge existing assessment techniques. For example, such problems may have reasonable alternative solutions, or conflicting objectives. Analyses of open-ended problems are often presented as free-form text since they require arguments and justifications for one solution over others, and students may differ in how they frame the problems according to their knowledge, beliefs and attitudes. This dissertation investigates how peer review may be used for formative assessment. Computer-Supported Peer Review in Education, a technology whose use is growing, has been shown to provide accurate summative assessment of student work, and peer feedback can indeed be helpful to students. A peer review process depends on the rubric that students use to assess and give feedback to each other. However, it is unclear how a rubric should be structured to produce feedback that is helpful to the student and at the same time to yield information that could be summarized for the instructor. The dissertation reports a study in which students wrote individual analyses of an open-ended legal problem, and then exchanged feedback using Comrade, a web application for peer review. The study compared two conditions: some students used a rubric that was relevant to legal argument in general (the domain-relevant rubric), while others used a rubric that addressed the conceptual issues embedded in the open-ended problem (the problem-specific rubric). While both rubric types yield peer ratings of student work that approximate the instructor's scores, feedback elicited by the domain-relevant rubric was redundant across its dimensions. On the contrary, peer ratings elicited by the problem-specific rubric distinguished among its dimensions. Hierarchical Bayesian models showed that ratings from both rubrics can be fit by pooling information across students, but only problem-specific ratings are fit better given information about distinct rubric dimensions.

No description available.

No description available.

While for many years two alternative approaches to building intelligent systems, symbolic AI and neural networks, have each demonstrated specific advantages and also revealed specific weaknesses, in recent years a number of researchers have sought methods of combining the two into a unified methodology which embodies the benefits of each while attenuating the disadvantages. This work sets out to identify the key ideas from each discipline and combine them into an architecture which would be practically scalable for very large network applications. The architecture is based on a relational database structure and forms the environment for an investigation into the necessary properties of a symbol encoding which will permit the singlepresentation learning of patterns and associations, the development of categories and features leading to robust generalisation and the seamless integration of a range of memory persistencies from short to long term. It is argued that if, as proposed by many proponents of symbolic AI, the symbol encoding must be causally related to its syntactic meaning, then it must also be mutable as the network learns and grows, adapting to the growing complexity of the relationships in which it is instantiated. Furthermore, it is argued that in order to create an efficient and coherent memory structure, the symbolic encoding itself must have an underlying structure which is not accessible symbolically; this structure would provide the framework permitting structurally sensitive processes to act upon symbols without explicit reference to their content. Such a structure must dictate how new symbols are created during normal operation. The network implementation proposed is based on K-from-N codes, which are shown to possess a number of desirable qualities and are well matched to the requirements of the symbol encoding. Several networks are developed and analysed to exploit these codes, based around a recurrent version of the non-holographic associati ve memory of Willshaw, et al. The simplest network is shown to have properties similar to those of a Hopfield network, but the storage capacity is shown to be greater, though at a cost of lower signal to noise ratio. Subsequent network additions break each K-from-N pattern into L subsets, each using D-from-N coding, creating cyclic patterns of period L. This step increases the capacity still further but at a cost of lower signal to noise ratio. The use of the network in associating pairs of input patterns with any given output pattern, an architectural requirement, is verified. The use of complex synaptic junctions is investigated as a means to increase storage capacity, to address the stability-plasticity dilemma and to implement the hierarchical aspects of the symbol encoding defined in the architecture. A wide range of options is developed which allow a number of key global parameters to be traded-off. One scheme is analysed and simulated. A final section examines some of the elements that need to be added to our current understanding of neural network-based reasoning systems to make general purpose intelligent systems possible. It is argued that the sections of this work represent pieces of the whole in this regard and that their integration will provide a sound basis for making such systems a reality.

No

No description available.

Recent trends in robotics have seen more general purpose robots being deployed in unstructured environments for prolonged periods of time. Such robots are expected to adapt to different environmental conditions, and ultimately take on a broader range of responsibilities, the specifications of which may change online after the robot has been deployed. We propose that in order for a robot to be generally capable in an online sense when it encounters a range of unknown tasks, it must have the ability to continually learn from a lifetime of experience. Key to this is the ability to generalise from experiences and form representations which facilitate faster learning of new tasks, as well as the transfer of knowledge between different situations. However, experience cannot be managed na¨ıvely: one does not want constantly expanding tables of data, but instead continually refined abstractions of the data – much like humans seem to abstract and organise knowledge. If this agent is active in the same, or similar, classes of environments for a prolonged period of time, it is provided with the opportunity to build abstract representations in order to simplify the learning of future tasks. The domain is a common structure underlying large families of tasks, and exploiting this affords the agent the potential to not only minimise relearning from scratch, but over time to build better models of the environment. We propose to learn such regularities from the environment, and extract the commonalities between tasks. This thesis aims to address the major question: what are the domain invariances which should be learnt by a long lived agent which encounters a range of different tasks? This question can be decomposed into three dimensions for learning invariances, based on perception, action and interaction. We present novel algorithms for dealing with each of these three factors. Firstly, how does the agent learn to represent the structure of the world? We focus here on learning inter-object relationships from depth information as a concise representation of the structure of the domain. To this end we introduce contact point networks as a topological abstraction of a scene, and present an algorithm based on support vector machine decision boundaries for extracting these from three dimensional point clouds obtained from the agent’s experience of a domain. By reducing the specific geometry of an environment into general skeletons based on contact between different objects, we can autonomously learn predicates describing spatial relationships. Secondly, how does the agent learn to acquire general domain knowledge? While the agent attempts new tasks, it requires a mechanism to control exploration, particularly when it has many courses of action available to it. To this end we draw on the fact that many local behaviours are common to different tasks. Identifying these amounts to learning “common sense” behavioural invariances across multiple tasks. This principle leads to our concept of action priors, which are defined as Dirichlet distributions over the action set of the agent. These are learnt from previous behaviours, and expressed as the prior probability of selecting each action in a state, and are used to guide the learning of novel tasks as an exploration policy within a reinforcement learning framework. Finally, how can the agent react online with sparse information? There are times when an agent is required to respond fast to some interactive setting, when it may have encountered similar tasks previously. To address this problem, we introduce the notion of types, being a latent class variable describing related problem instances. The agent is required to learn, identify and respond to these different types in online interactive scenarios. We then introduce Bayesian policy reuse as an algorithm that involves maintaining beliefs over the current task instance, updating these from sparse signals, and selecting and instantiating an optimal response from a behaviour library. This thesis therefore makes the following contributions. We provide the first algorithm for autonomously learning spatial relationships between objects from point cloud data. We then provide an algorithm for extracting action priors from a set of policies, and show that considerable gains in speed can be achieved in learning subsequent tasks over learning from scratch, particularly in reducing the initial losses associated with unguided exploration. Additionally, we demonstrate how these action priors allow for safe exploration, feature selection, and a method for analysing and advising other agents’ movement through a domain. Finally, we introduce Bayesian policy reuse which allows an agent to quickly draw on a library of policies and instantiate the correct one, enabling rapid online responses to adversarial conditions.

Both semantic web applications and individuals are in need of knowledge infrastructures that can be used in dynamic and distributed environments where autonomous entities create knowledge and build their own view of a domain. The prevailing view today is that the process of ontology evolution is difficult to monitor and control, so few efforts have been made to support such a controlled process formally involving several ontologies. The new paradigm we propose is to use an infrastructure mechanism that processes ontology change proposals from autonomous entities while maintaining user-defined consistency between the ontologies of these entities. This makes so called semantic autonomy possible. A core invention of our approach is to formalise consistency constraints as so called spheres of consistency that define 1) knowledge regions within which consistency is maintained and 2) a variable degree of proof-bounded consistency within these regions. Our infrastructure formalism defines a protocol and its computational semantics, as well as a model theory and proof theory for the reasoning layer of the mechanism. The conclusion of this thesis is that this new paradigm is possible and beneficial, assuming that the knowledge representation is kept simple, the ontology evolution operations are kept simple and one proposal is processed at a time.

To improve the accuracy in capturing the motion of deformable objects, a reconfigurable multi-camera active-vision system which can dynamically reposition its cameras online is proposed, and a design for such a system, along with a methodology to select the near-optimal positions and orientations of the set of cameras, is presented. The active-vision system accounts for the deformation of the object-of-interest by tracking triangulated vertices in order to predict the shape of the object at subsequent demand instants. It then selects a system configuration that minimizes the expected error in the recovered position of each of these vertices. Extensive simulations and experiments have verified that using the proposed reconfigurable system to both translate and rotate cameras to near-optimal poses is tangibly superior to using cameras which are either static, or can only rotate, in minimizing the error in recovered vertex positions.