Deep learning for computer vision: a comparison between convolutional neural networks and hierarchical temporal memories on object recognition tasks

In recent years, Deep Learning techniques have shown to perform well on a large variety of problems both in Computer Vision and Natural Language Processing, reaching and often surpassing the state of the art on many tasks. The rise of deep learning is also revolutionizing the entire field of Machine Learning and Pattern Recognition pushing forward the concepts of automatic feature extraction and unsupervised learning in general. However, despite the strong success both in science and business, deep learning has its own limitations. It is often questioned if such techniques are only some kind of brute-force statistical approaches and if they can only work in the context of High Performance Computing with tons of data. Another important question is whether they are really biologically inspired, as claimed in certain cases, and if they can scale well in terms of "intelligence". The dissertation is focused on trying to answer these key questions in the context of Computer Vision and, in particular, Object Recognition, a task that has been heavily revolutionized by recent advances in the field. Practically speaking, these answers are based on an exhaustive comparison between two, very different, deep learning techniques on the aforementioned task: Convolutional Neural Network (CNN) and Hierarchical Temporal memory (HTM). They stand for two different approaches and points of view within the big hat of deep learning and are the best choices to understand and point out strengths and weaknesses of each of them. CNN is considered one of the most classic and powerful supervised methods used today in machine learning and pattern recognition, especially in object recognition. CNNs are well received and accepted by the scientific community and are already deployed in large corporation like Google and Facebook for solving face recognition and image auto-tagging problems. HTM, on the other hand, is known as a new emerging paradigm and a new meanly-unsupervised method, that is more biologically inspired. It tries to gain more insights from the computational neuroscience community in order to incorporate concepts like time, context and attention during the learning process which are typical of the human brain. In the end, the thesis is supposed to prove that in certain cases, with a lower quantity of data, HTM can outperform CNN.

Identifying specific interpretations and use of safety behaviours in people with distressing visual hallucinations: an exploratory study

Visual hallucinations (VH) are a common experience and can be distressing and disabling, particularly for people suffering from psychotic illness. However, not everyone with visual hallucinations reports the experience to be distressing. Models of VH propose that appraisals of VH as a threat to wellbeing and the use of safety seeking behaviours help maintain the distress.

Exploratory studies on coordination chemistry of a redox-active bridging ligand: synthesis, properties and solid state structures of the complexes

The explorative coordination chemistry of the bridging ligand TTF-PPB is presented. Its strong binding ability to Co(II) and then to Ni(II) or Cu(II) in the presence of hexafluoroacetylacetonate (hfac(-)), forming new mono-and dinuclear complexes 1-3, is described. X-ray crystallographic studies have been conducted in the case of the free ligand TTF-PPB as well as its complexes [Co(TTF-PPB)(hfac)(2)] (1) and [Co(hfac)(2)(mu-TTF-PPB)Ni(hfac)(2)] (2). Each metal ion is bonded to two bidentate hfac-anions through their oxygen atoms and two nitrogen atoms of the PPB moiety with a distorted octahedral coordination geometry. Specifically, nitrogen donor atoms of TTF-PPB adopt a cis-coordination but not in the equatorial plane, which is quite rare. Electronic absorption, photoinduced intraligand charge transfer ((1)ILCT), and electrochemical behaviour of 1-3 have been investigated. UV-Vis spectroscopy shows very strong bands in the UV region consistent with ligand centred pi-pi* transitions and an intense broad band in the visible region corresponding to a spin-allowed pi-pi* (1)ILCT transition. Upon coordination, the (1)ILCT band is bathochromically shifted by 3100, 6100 and 5900 cm(-1) on going from 1 to 3. The electrochemical studies reveal that all of them undergo two reversible oxidation and one reversible reduction processes, ascribed to the successive oxidations of the TTF moiety and the reduction of the PPB unit, respectively.

Aquatic therapies in patients with compromised left ventricular function and heart failure

With water immersion, gravity is partly eliminated, and the water exerts a pressure on the body surface. Consequently there is a blood volume shift from the periphery to the central circulation, resulting in marked volume loading of the thorax and heart. This paper presents a selection of published literature on water immersion, balneotherapy, aqua exercises, and swimming, in patients with left ventricular dysfunction (LVD) and/or stable chronic heart failure (CHF). Based on exploratory studies, central hemodynamic and neurohumoral responses of aquatic therapies will be illustrated. Major findings are: 1. In LVD and CHF, a positive effect of therapeutic warm-water tub bathing has been observed, which is assumed to be from afterload reduction due to peripheral vasodilatation caused by the warm water. 2. In coronary patients with LVD, at low-level water cycling the heart is working more efficiently than at lowlevel cycling outside of water. 3. In patients with previous extensive myocardial infarction, upright immersion to the neck resulted in temporary pathological increases in mean pulmonary artery pressure (mPAP) and mean pulmonary capillary pressures (mPCP). 4. Additionally, during slow swimming (20-25m/min) the mPAP and/or PCP were higher than during supine cycling outside water at a 100W load. 5. In CHF patients, neck- deep immersion resulted in a decrease or no change in stroke volume. 6. Although patients are hemodynamically compromised, they usually maintain a feeling of well-being during aquatic therapy. Based on these findings, clinical indications for aquatic therapies are proposed and ideas are presented to provoke further research.

The relationship between pain intensity and severity and depression in older people: exploratory study

BACKGROUND: Pain and depression are known to be associated in later life, and both have a negative effect on physical performance both separately and in combination. The nature of the relationships between pain intensity and depression in elderly persons experiencing pain is less clear. The objectives of this study were to explore which factors are associated with depressed mood in older people experiencing pain, and to test the hypothesis that older people experiencing pain are at risk of depressed mood according to the severity or frequency of their pain. In addition we explored whether other potentially modifiable factors might increase the risk of depressed mood in these persons. METHODS: The study is a secondary analysis of baseline data for four hundred and six community-dwelling non-disabled people aged 65 and over registered with three group practices in suburban London who had experienced pain in the past 4 weeks. Intensity and frequency of pain was measured using 24 item Geriatric Pain Measure (GPM) and the presence of depressive symptoms using the 5 item Mental Health Inventory. Risk for social isolation was measured using the 6 item Lubben Social Network scale and instrumental activities of daily living (IADL) were also measured. RESULTS: Overall 76 (19%) had depressed mood. Pain frequency and severity were not statistically significantly associated with depressed mood in this population. In multivariate analyses, significant predictors of the presence of depressive symptoms were difficulties with basic ADLs (OR 2.8, 95% CI 1.1.7.8), risk for social isolation (OR 4.1, 95% CI 1.8-9.3), and basic education only (OR 2.2, 95% CI 1.1-4.4). CONCLUSION: Older people experiencing pain are also likely to experience depression. Among those experiencing pain, social network and functional status seem to be more important predictors of depressive symptoms than the severity of pain. Further studies should evaluate whether improvement of social network and functional status might reduce depressive symptoms in older patients.