We study the creation of physical adversarial examples, which are robust to real-world transformations, using a limited number of queries to the target black-box neural networks. We observe that robust models tend to be especially susceptible to foreground manipulations, which motivates our novel Foreground attack. We demonstrate that gradient priors are a useful signal for black-box attacks and therefore introduce an improved version of the popular SimBA. We also propose an algorithm for transferable attacks that selects the most similar surrogates to the target model. Our black-box attacks outperform state-of-the-art approaches they are based on and support our belief that the concept of model similarity could be leveraged to build strong attacks in a limited-information setting.

A traditional helicopter requires 60 - 80% more power to hover than when in forward or lateral flight, making the manoeuvre extremely power inefficient. To maximise efficiency, industrially many properties of the helicopter and rotor have been changed and tested, for example: optimising blade shape, fuselage shape and changing weights of different helicopter components. This report in particular aims to find a relationship between power efficiency and angular velocity for a twin bladed hovering helicopter with a single rotor. The angular velocity of a blade measures the frequency of its revolution about a fixed point. A theoretical approach was first taken and then justified with empirical data. Firstly, a model for power efficiency was derived with William Froude’s momentum and blade element theory. The efficiency equations incorporated the thrust and power coefficients. Therefore, the research focused on determining values for these coefficients by manipulating equations, using industrial specifications and simulations from the XFOIL software. In order to validate the accuracy for such theoretically generated data, an experiment was conducted for a comparison. The theoretical and empirical data were concurrent, as they followed a similar trend and the empirical values overlapped within the theoretical error bars. The power efficiency for different angular velocities were then found by substituting values for the coefficients. The results demonstrated a positive relationship; where, as angular velocity increases, power efficiency increases too, then plateaus and repeats the same trend once again. The research raises many questions and could be extended by determining if a similar relationship exists for tri-copters and quadcopters.

Triage is the process by which nurses manage hospital emergency departments by assigning patients varying degrees of urgency. While triage algorithms such as the Emergency Severity Index (ESI) have been standardized worldwide, many of them are highly inconsistent, which could endanger the lives of thousands of patients. One way to improve on nurses’ accuracy is to use machine learning models (ML), which can learn from past data to make predictions. We tested six ML models: random forest, XGBoost, logistic regression, support vector machines, k-nearest neighbors, and multilayer perceptron. These models were tasked with predicting whether a patient would be admitted to the intensive care unit (ICU), another unit in the hospital, or be discharged. After training on data from more than 30,000 patients and testing using 10-fold cross-validation, we found that all six models outperformed ESI. Of the six, the random forest model achieved the highest average accuracy in predicting both ICU admission (81% vs. 69% using ESI; p<0.001) and hospitalization (75% vs. 57%; p<0.001). These models were then added to an Android application, which would accept patient data, predict their triage, and then add them to a priority-ordered waiting list. This approach may offer significant advantages over conventional triage: mainly, it has a higher accuracy than nurses and returns predictions instantaneously. It could also stand-in for triage nurses entirely in disasters, where medical personnel must deal with a large influx of patients in a short amount of time.

The use of brine shrimp nauplii to test for the overall toxicity of sediment samples is proposed. Brine shrimp nauplii were cultured with different concentrations of heavy metals, including chromium (III), copper (II), nickel, lead and zinc, and organic pollutants, including triclosan, oxybenzone, octinoxate and bisphenol A. The brine shrimp nauplii were observed under a dissection microscope to determine the death rate. Results showed that brine shrimp nauplii are more sensitive to copper, cadmium, bisphenol A and oxybenzone. The LC50 (24h) are 55.5, 24.9, 5.6 and 2.7 ppm respectively. Zinc is likely to have synergistic toxic effect with nickel or lead. The synergistic toxic effects of other heavy metals and organic pollutants should be confirmed with further investigations. Brine shrimp nauplii were treated with extracts from sediment samples collected from the oyster culture zone of the Deep Bay, namely Pak Nei, Sha Kiu Tsuen and Hang Hau Tsuen. The sediment samples were extracted with neutral sodium acetate to dissolve the exchangeable heavy metal ions and some organic pollutants. The death rate of brine shrimp nauplii treated with the sediment extract of Hang Hau Tsuen was similar to 1 ppm PBA. It was also about 10 to 20% higher than that of the other two sites (Pak Nei and Sha Kiu Tsuen). Since Hang Hau Tsuen is closer to the residential area and Lau Fau Shan Seafood Market than the other two sites, its sediment sample is likely to have a higher level of environmental pollutants. The results suggest that brine shrimp nauplii may be used as a biomarker to monitor the environmental changes in the overall level of pollutants in sediment samples.

The lack of drinking water in human settlements triggers a series of problems that are linked and affect the development of humanity: health problems, lack of water security for companies, lack of jobs, insecurity, among others. We observe this problem in the communities of the municipality of Las Vigas de Ramírez, Veracruz, where there is a great problem with the water supply, although there is a high presence of mist. Faced with this situation, we undertook the task of investigating a water harvesting method that is easy to implement, operate and maintain. We investigated and analyzed the methods of mist condensation through physical barriers, finding that the polyethylene shadow mesh was the means to achieve this, because it allows the passage of the wind, it is very light, easy to manipulate and above all that it presents the phenomenon of percolation that allows water droplets of various diameters to be accommodated therein. We designed a device that allows to present a mist catchment area through a prismatic structure enabled with meshes and condensed water receivers, portable, easy to use and maintenance and very economical with a performance of 20 liters per day. To achieve our project, factors such as air humidity, dew point, wind speed and direction, height, temperatures and available spaces must be considered.

This project aimed to study the motion which occurred from the end point on the circumference of the outermost circle by moving the center on the circumference of a preceding circle and the center of an innermost circle at origin. According to the study, when angular velocity was changed, it caused the different of loci. Based on the above information, finding the locus of the point on circumference of n-th circle that formed by moving the center of any radius circles on circumference of preceding set of circles was studied to get general equation. A set of circle and locus were created with GSP program. First, set the same radius circles on the X-axis with the first circle at origin, then found the relationship that occurred from the characteristics of locus. The result showed that if the ratios of angular velocity are 1:1:1, 2:2:2, 3:3:3, ..., …, n:n:n or 1:2:3, 2:4:6, 3:6:9, …,nw1:nw2:nw3, the characteristics of locus will be the same, while the others will be different. Finally, the equation of locus was found as follow: (x,y) = { ..........see in abstract...........} when .........see in abstract........... Where ri is the radius of i-th circle, zeta i is an angle between the radius of i-th circle and X-axis, wi is the angular velocity, t is elapsed time and alpha i is a starting angle between the radius of i-th circle and X-axis.

Two years ago, I attended a robot contest, in which one of the missions required the robot to follow the pedestrian to complete the task. At that time, I used their demo program to complete the task. Not long after, I found two main issues: 1. The program follows the closest point read by the depth camera, which if I walk close to a wall next to, the robot may likely ‘follow’ the wall. 2. Not to mention if another pedestrian crosses between the robot and the target. Regarding these two issues, I decided to improve it. We’ve designed a procedure of using YOLO Object Detection and Person re-identification to re-identify the target for continuous following.

Inhibitory effects of the secondary metabolite of actinomycete were examined on cell cycle of the yeasts of S. pombe and S. cerevisiae. The secondary metabolite was obtained from cultivation of the actinomycete isolated from the soil of Owakudani in Hakone, Japan. The fifth fraction of the secondary metabolite by ODS column separation (HK-T5), which was soluble to pure methanol, was used in the present experiments. The HK-T5 brought about the delay of forming colonies of S. pombe for about 11 days compared to that cultivated without the HK-T5. The delay of the colony formation was longer for the S. pombe cultivated with more amount of the HK-T5. The cultivation with HK-T5 also brought about the extension of the lifespan of the S. pombe for more than 10 weeks in a liquidus medium. The cell life recovered the ordinary manner by removal of the HK-T5, meaning that the activities of the HK-T5 is reversible. These facts confirm the suppression of cell cycle, and the delay of cell growth by the HK-T5. These phenomena were similarly observed for S. cerevisiae. Comparison of the action of HK-T5 with hydroxyurea, which is an anticancer drug inhibiting the cell cycle at S phase, clarified that the inhibitory action of HK-T5 worked at the phase earlier than S phase. The combined effects of HK-T5 on the cell cycle were evaluated with triamcinolone acetonide (TA), or aspirin, the former of which is a drug synchronizing cancer cells in S phase, and the latter keeping human cells in G1/G0 phases. The combined use of HK-T5 with TA synchronized the cells at the phase slightly proceeding from G1 to S phase without toxicity. On the other hand, the combined use with aspirin made the inhibitory effect of HK-T5 inactive. Hence, the HK-T5 is attractive as a drug for the extension of cell lifespan, and anticancer therapy.

This work is devoted to solving the problem of orientation in the space of visually impaired people. Working on the project, a new way of transmitting visual information through an acoustic channel was invented. In addition, was developed the device, which uses distance sensors to analyze the situation around a user. Thanks to the invented algorithm of transformation of the information about the position of the obstacle into the sound of a certain tone and intensity, this device allows the user to transmit subject-spatial information in real time. Currently, the device should use a facette locator made of 36 ultrasonic locators grouped in 12 sectors by the azimuth and 3 spatial cones by the angle. Data obtained in such a way is converted into its own note according to the following pattern : the angle of the place corresponds to octave, the azimuth corresponds to the note and the distance corresponds to the volume. The choice of the notes is not unambiguous. However, we used them for the reason that over the centuries, notes have had a felicitous way of layout on the frequency range and on the logarithmic scale. Therefore, the appearance of a new note in the total signal will not be muffled by a combination of other notes. Consequently, a blind person, moving around the room with the help of the tone and volume of the sound signals, will be able to assess the presence and location of all dangerous obstacles. After theoretical substantiation of the hypothesis and analysis of the available information, we started the production of prototypes of the devices that would implement the idea of transmitting information via the acoustic channel.

Double Parallelogram Lifting Mechanism (DPLM) is a compact and stable lifting mechanism with a large extension range widely adopted in robot designs. Rubber bands and springs are often installed on the DPLM to lighten the motors' load and maintain its height, yet the installation positions are often obtained through trial and error. This project aims at finding the optimal rubber band installation positions for DPLM using modeling and optimization techniques. A mathematical model which describes the forces and moments acting on all the linkages of DPLM was derived based on the conditions for the static equilibrium and verified with a 3D simulation software. A genetic algorithm (GA) was implemented to optimize rubber band installation positions, which managed to find solutions with the overall root-mean-square- error (RMSE) of the net moment less than 2 for 2 to 6 rubber bands. A further statistical analysis of 50000 random rubber band samples showed that installing rubber bands in triangles is the best solution with the overall lowest RMSE. A test was conducted with a prototype of the DPLM and the results were consistent with our model and optimization. This project derived and verified a mathematical model for the DPLM, and found the optimal way and positions to install rubber bands. The results of this project provides a theoretical basis for controlling DPLM with rubber bands, allowing it to be further adopted in industrial robots that require repetitive lifting and lowering such as inspection robots and aerial work platforms.

Stochastic gradient descent (SGD) is one of the preferred online optimization algorithms. However, one of its major drawbacks is its predisposition to forgetting previous data when optimizing through a data stream, also known as catastrophic interference. In this project, we attempt to mitigate this drawback by proposing a new low-cost approach which incorporates backward queried images with SGD during online training. Under this new approach, we propose that for every new training sample through the data stream, the neural network is optimized using the corresponding backward queried image from the initial dataset. After compiling the accuracy of the proposed method and SGD under a data-stream of 50,000 training cases with 10,000 test cases and comparing our algorithm to SGD, we see substantial improvements in the performance of the neural network with two different MNIST datasets (Fashion and Kuzushiji), classifying the MNIST datasets at a high accuracy for the mean, minimum, lower quartile, median, and upper quartile, while maintaining lower standard deviation in performance, demonstrating that our proposed algorithm can be a potential alternative to online SGD.