Search

Sedentary behavior is considered as a major public health challenge, linked with many chronic diseases and premature mortality. In this paper, we propose a steps counting -based machine learning approach for the prediction of sedentary behavior. Our work focuses on analyzing historical data from multiple users of wearable physical activity trackers and exploring the performance of four machine learning algorithms, i.e., Logistic Regression, Random Forest, XGBoost, Convolutional Neural Networks, as well as a Majority Vote Ensemble of the algorithms. To train and test our models we employed a crowd sourced dataset containing a month's data of 33 users. For further evaluation, we employed a dataset containing 6 months of data of an additional user. The results revealed that while all models succeed in predicting next-day sedentary behavior, the ensemble model outperforms all baselines, as it manages to predict sedentary behavior and reduce false positives more effectively. On the multi-subjects test dataset, our ensemble model achieved an accuracy of 82.12% with a sensitivity of 74.53% and a specificity of 85.71%. On the additional unseen dataset, we achieved 76.88% in accuracy, 63.27% in sensitivity and 81.75% in specificity. These outcomes provide the ground towards the development of real-life artificially intelligent systems for sedentary behavior prediction.

We present two methods for investigating the role of familiarity in processing kinetic-depth-effect signals produced by rotating concave objects. Motion-parallax signals recover the true three-dimensional shape of concave objects and they compete against the bias to perceive objects as convex, rather than concave. We test the hypothesis that the cognitive influence of object familiarity favors the bias for convexity, thus strengthening the illusion of convexity for familiar objects. Data with physical and virtual objects obtained with healthy controls confirm this trend and provide a baseline for comparison with schizophrenia patients.

The link between cardiovascular disease (CVD) and air pollution risk factors is not fully understood and has not been scientifically investigated in many developing Asian countries such as Vietnam and the Philippines. The aim of this paper is to investigate practising cardiologists' opinions and knowledge about air pollution exposure potentially triggering cardiovascular disease in those countries. The results of the study are of importance to not only avoid the burden of CVD in term of mortality and morbidity with all their ensuing costs, but also to understand physicians' evaluations of the environmental risk factors to CVD and to suggest measures to overcome any shortcomings.

The air pollution situation in Vietnam and the Philippines has deteriorated in recent years, and mortality rates from cardiovascular diseases (CVD) have remained in the top ten of all causes of deaths for the last few years. The aim of this paper is to understand and model the awareness of cardiologists of the potential impact of air pollution on CVD in both nations. A full-scale survey covering 321 cardiologists was conducted in Vietnam (27 hospitals across the country) and the Philippines (members of the Philippine Heart Association with a subset from Davao Province). The paper reports on extensive results obtained through logistic regression modelling.

Individuals with schizophrenia (SZ) perceive various visual stimuli differently than healthy controls. This is especially true of the “hollow-face illusion.” We have created a virtual rendering system that exploits this well-known fact. We also explain the addition of a head-tracking feature that provides a more realistic spatio-temporal interaction for our virtual-reality apparatus. Our system is capable of quantifying the amount of time a person perceives the illusion and affords a comparison between SZ patients and controls, thus offering a diagnostic tool for SZ.

This paper presents a systematic application of machine learning techniques for classifying high-density EEG signals elicited by face and non-face stimuli. The two stimuli used here are derived from the vase-faces illusion and share the same defining contours, differing only slightly in stimulus space. This emphasizes activity differences related to high-level percepts rather than low-level attributes. This design decision results in a difficult classification task for the ensuing EEG signals. Traditionally, EEG analyses are done on the basis of signal processing techniques involving multiple instance averaging and then a manual examination to detect differentiating components. The present study constitutes an agnostic effort based on purely statistical estimates of three major classifiers: L1-norm logistic regression, group lasso and k Nearest Neighbors (kNN); kNN produced the worst results. L1 regression and group lasso show significantly better performance, while being abl e to identify distinct spatio-temporal signatures. Both L1 regression and group lasso assert the saliency of samples in 170ms, 250ms, 400ms and 600ms after stimulus onset, congruent with the previously reported ERP components associated with face perception. Similarly, spatial locations of salient markers point to the occipital and temporal brain regions, previously implicated in visual object perception. The overall approach presented here can provide a principled way of identifying EEG correlates of other perceptual/cognitive tasks.

Embedded passives account for a very large part of today's electronic assemblies. This is particularly true for products such as cellular phones, camcorders, computers and several critical defense devices. Market pressures for new products with more features, smaller size and lower cost demand smaller, compact, simpler substrates. An obvious strategy is to reduce the number of surface mounted passives by embedding them in the substrate. In addition, current interconnect technology to accommodate surface mounted passives imposes certain limits on board design which constrain the overall system speed. Embedding passives is one way to minimize the functional footprint while at the same time improving performance. This paper discusses thin film technology based on resin coated copper capacitive (RC3) nanocomposites. In particular, we highlight recent developments on high capacitance, large area, thin film passives and their integration in System in a Package (SiP). A variety of RC3 nanocomposite thin films ranging from 8 microns to 50 microns thick were processed on Cu substrates by liquid coating. Multilayer embedded capacitors resulted in high capacitance 16–28 nF. The fabricated test vehicle also included two embedded resistor layers with resistance in the range of 15 ohms to 100 kohms. To enable high performance devices, an embedded resistor must meet certain tolerances. The embedded resistors can be laser trimmed to a tolerance of ≪5%, which is usually acceptable for most applications. We have an extended embedded passives solution that has been demonstrated both through its high wireability designs and package performance to be perfectly suited for the system in package (SiP) applications. As a case study, we have designed and fabricated eight layer high density internal passive core and subsequently applied fine geometry 3 buildup layers to form a 3-8-3 structure. The passive core technology is capable of providing up to 6 layers of embedded capacitance and could be extended further. This effort is an integrated approach centering on three interrelated fronts: (1) materials development and characterization; (2) fabrication, and (3) integration at the device level.

This paper examines the use of nanocomposites in the area of “green” technology. A variety of green materials for advanced organic packaging have been developed. These include capacitors and resistors as embedded passives, resin coated Cu (RCC) as buildup layers, highly conducting nano-micro media for Z-interconnects, lead free assembly paste, ZnO based additives and magnetic materials. Nanocomposites can provide high capacitance densities, ranging from 5 nf/inch2 to 25 nF/inch2, depending on composition, particle size and film thickness. The electrical properties of capacitors fabricated from BaTiO 3 -epoxy nanocomposites showed a stable capacitance over a temperature range from 20°C to 120°C. A variety of printable discrete resistors with different sheet resistances, ranging from 1 ohm to 120 Mohm, processed utilizing a large panel format (19.5 × 24 inches) have been fabricated. Low resistivity nanocomposites, with volume resistivity in the range of 10−4 ohm-cm to 10−6 ohm-cm depending on composition, particle size, and loading can be used as conductive joints for high frequency and high density interconnect applications. A variety of metals including Cu, Ag, LMP (low melting point) and LMP-coated Cu fillers have been used to make halogen free, lead free electrically conducting adhesive technology as an alternative to solders. Halogen free resin modified with ceramics/organic particles can produce low Dk resin coated Cu (RCC) with Dk value in the range between 4.2 and 3.2. Similarly, low loss RCC materials can be produced by combining HF resin with low loss fillers. The mechanical strength of the various RCC was characterized by a 90 degree peel test and measurement of tensile strength. RCC exhibited peel strength with Gould's JTC-treated Cu as high as 6 lbs/inch for halogen free RCC. These halogen free RCC materials exhibit coefficients of thermal expansion (CTE), ranging from 27 ppm/°C to 32ppm/°C. Altogether, this is a new direction in the development of Green Packages and more specifically in the development of coreless substrates for semiconductor packaging.

Jain et al. [1] showed that a strong motion signal in one modality (visual/auditory) influences the perception of a simultaneously present weak motion signal in the other modality (auditory/visual) for motion along the three cardinal axes. Furthermore, they also observed auditory crossmodal motion aftereffects (MAE) for all directions and visual crossmodal MAE for vertical direction of motion. We developed a neurophysiologically plausible computation model to explain the observed interactions. The current model was based on the hypothesis that these crossmodal interactions are mediated by the higher integrative multimodal areas, such as the superior colliculus (SC), through feedback connections to the lower unimodal areas in the human brain.

Brominated flame retardants, namely polybrominated diphenyl ethers (PBDEs), were one of the main materials used to reduce the flammability of consumer goods and electronics. Growing evidence shows that PBDE compounds are making their way into the environment. These chemicals may cause health effects, prompting many nations to ban or suspend their use in new consumer goods.

This paper discusses thin film technology based on resin coated copper capacitive (RC3) nanocomposites. In particular, we highlight recent developments on high capacitance, large area, thin film passives, their integration in printed wiring boards (PWB), system in package (SiP) and chip package substrates and the reliability of the embedded capacitors. A variety of RC3 nanocomposite thin films ranging from 2 microns to 50 microns thick were processed on PWB substrates by liquid coating or printing processes. SEM micrographs showed uniform particle distribution in the coatings. The electrical performance of composites was characterized by dielectric constant (Dk), capacitance and dissipation factor (loss) measurements. Nanocomposites resulted in high capacitance density (7-500 nF/inch2) at 1 MHz. The manufacturability of these films and their reliability has been tested using large area (13 inch times 18 inch or 19.5 inch times24 inch) test vehicles. Reliability of the RC3 nanocomposite was ascertained by IR-reflow, PCT (pressure cooker test) and solder shock. As a case study, an example of RC3 based multilayer embedded capacitor construction for a flip-chip plastic ball grid array package with a 300 mum core via pitch is given. This effort is an integrated approach centering on three interrelated fronts: (1) materials development and characterization; (2) fabrication, and (3) integration at the device level.

This paper discusses micro-filled epoxy-based conducting adhesives modified with nanoparticles, conducting polymers, and low melting point (LMP) fillers for z-axis interconnections, especially as they relate to package level fabrication, integration, and reliability. A variety of conducting adhesives with particle sizes ranging from 80 nm to 15 mum were incorporated as interconnects in printed wiring board (PWB) or laminated chip carrier (LCC) substrates. SEM and optical microscopy were used to investigate the micro-structure, and conducting and sintering mechanisms. Volume resistivity of modified adhesives is in the range of 10-5 to 10-6 ohm-cm. Adhesives formulated with a conducting polymer exhibited tensile strength with Gould's JTC-treated Cu ges 3800 PSI, and as low as 1800 PSI for a conducting polymer-LMP based system. There was no delamination of conductive joints after 3X IR-reflow, pressure cooker test (PCT), and solder shock. Among all, the conducting polymer modified micro-filled adhesives showed the highest mechanical strength. The paper also describes a combinatorial approach to the synthesis of LMP coated particles. Several conductive adhesives were used in a z-axis interconnect construction for a laminate chip carrier and printed wiring board (PWB). The present process allows fabrication of z-interconnect conductive joints having diameters in the range of 55-300 microns. The processes and materials used to achieve smaller feature dimensions, satisfy stringent registration requirements, and achieve robust electrical interconnections are discussed.

This paper examines the use of nanocomposites or materials in the area of printing technology. A variety of printable nanomaterials for advanced organic packaging have been developed. This includes nano capacitors and resistors as embedded passives, nano magnetic materials, multifunctional materials, etc. Nanocomposites can provide high capacitance densities, ranging from 5 nf/inch2 to 25 nF/inch2, depending on composition, particle size and film thickness. The electrical properties of capacitors fabricated from BaTiO3-epoxy nanocomposites showed a stable capacitance and low loss over a temperature range from 25degC to 100degC. A variety of printable discrete resistors with different sheet resistances, ranging from 1 ohm to 120 Mohm, processed on large panels (19.5 inches times 24 inches) have been fabricated. Low resistivity nanocomposites, with volume resistivity in the range of 10-4 ohm-cm to 10-6 ohm-cm depending on composition, particle size, and loading can be used as conductive joints for high frequency and high density interconnect applications. Thermosetting polymers modified with ceramics can produce low k dielectrics with k value in the range between 5.41 and 3.59. Similarly, low loss dielectric materials can be produced form mixing epoxy with silica or other low loss fillers. Reliability of the nanocomposites was ascertained by IR-reflow, thermal cycling, pressure cooker test (PCT), and solder shock. Change in capacitance after 3X IR-reflow and after 1000 cycles of deep thermal cycling (DTC) between -55degC and 125degC was within 5%. Most of the nanocomposites in the test vehicle were stable after IR-reflow, PCT, and solder shock.

We developed a neurocomputational model for the integration of visual and auditory stimuli. The model comprises three main stages. The first visual stage is motion extraction, with a corresponding stage for auditory processing. The second stage models visual-auditory neural interactions, simulating a neural network involving the superior colliculus. The third stage is a global integration stage, simulating higher cortical areas. Simulation results with this model agree closely with experimental data.

The potential of lossy image compression is investigated in a well defined task domain, specifically, clinical assessment of chromatic, surgical and hematopathology specimens. Two criteria were employed, just noticeable difference and largest tolerable distortion. Compression tolerances differed among pathologists, but conformed to well defined upper and lower limits. The level of tolerable compression was significantly larger for the second criterion. It is concluded that lossy image compression is feasible for diagnostic pathology and may hold promise for telepathology applications.

A computational model is presented, the results of which agree closely with those from experiments in human motion perception. Both Fourier and non-Fourier motion components are included. This model consists of two parallel pathways: a first-order pathway and a second-order pathway that uses full-wave rectification. The model predicts behavioral performance accurately and offers evidence for the feasibility of a dual pathway in biological motion processing.

Humans can perceive non-Fourier visual motion, i.e., motion that contains no coherent structure in the distribution of stimulus energy in the spatiotemporal frequency domain. The question that still remains unanswered is whether this non-Fourier motion is analyzed by a separate neural pathway or whether there is a unique pathway that is responsible for both Fourier and non-Fourier motion analysis. We present experimental evidence that strongly supports the separate-pathway hypothesis, and we suggest a possible computational model that can account for the data.

We have developed a new computational model for texture perception which is physiologically plausible and mimics human performance. Our model tries to simulate the visual processing characteristics by incorporating mechanisms tuned to luminance, orientation, spatial-frequency and color, which are characteristic features of any textural image. We obtained a very good correlation between human performance and our model simulations with various strategic texture patterns. The highlights of our model are incorporation of chromatic mechanisms to treat color images, in addition to grey-level ones, and the extension of the concept of double-opponency beyond color. The model could be utilized in the area of image processing, machine vision and pattern recognition, and scientific visualization. >

Classes of powerful stimuli that reliably elicit wrong depth percepts are examined. The depth percepts in turn cause the stimuli to appear to move vividly as observers move in front of them. These illusory percepts highlight the issue of how we obtain a stable percept of the environment as we move about, despite considerable changes in the retinal images. Experimental results on the role of pictorial cues, and possible explanations for the illusory motion are presented. The interaction of data-driven mechanisms and schema-driven cognitive processes is examined.

As mobile computing and telecommunication electronics are spreading fast, a demand for microelectronics packaging schemes for high density, fine pitch, high performance and low cost becomes even more severe. Specifically, the conventional printed circuit board (PCB) with plated-through-hole (PTH) vias is difficult to justify its manufacturing cost as well as the packaging density requirement. To meet this challenge, several new manufacturing schemes of high density and high performance PCB have been recently introduced such as surface laminar circuit (SLC), array layer inner via hole (ALIVH) and others. This new PCB fabrication process is categorized as sequential build-up (SBU) or build-up multilayer (BUM) using laminate-based substrates, where via holes are filled with a conductive paste material to make reliable vertical or Z-interconnects. In this paper, a new electrically conducting paste material to be used for via filling is introduced. The new conducting material consists of a conducting filler powder coated with a low melting point metal or alloy, a mixture of several thermoset resins, and other minor organic additives. By varying the filler content and resin chemistry, several formulations have been produced to fill via holes with a high aspect ratio. Via fill experiments have been performed to demonstrate void-free microstructure with good electrical continuity. Various bulk properties such as thermal, electrical and mechanical have also been characterized to understand the material behavior during via filling as well as the field service.

The Bayesian relevance feedback approach introduced with the PicHunter system (Cox, Miller, Omohundro and Yianilos, Int. Conf. on Pattern Recognition, pp. 361-369, 1996) is extended to include hidden semantic attributes. The general approach is motivated and experimental results are presented that demonstrate significant reductions in search times (28-32%) using these annotations