Embedding and extraction of knowledge in tree ensemble classifiers

The embedding and extraction of knowledge is a recent trend in machine learning applications, e.g., to supplement training datasets that are small. Whilst, as the increasing use of machine learning models in security-critical applications, the embedding and extraction of malicious knowledge are equivalent to the notorious backdoor attack and defence, respectively. This paper studies the embedding and extraction of knowledge in tree ensemble classifiers, and focuses on knowledge expressible with a generic form of Boolean formulas, e.g., point-wise robustness and backdoor attacks. For the embedding, it is required to be preservative (the original performance of the classifier is preserved), verifiable (the knowledge can be attested), and stealthy (the embedding cannot be easily detected). To facilitate this, we propose two novel, and effective embedding algorithms, one of which is for black-box settings and the other for white-box settings. The embedding can be done in PTIME. Beyond the embedding, we develop an algorithm to extract the embedded knowledge, by reducing the problem to be solvable with an SMT (satisfiability modulo theories) solver. While this novel algorithm can successfully extract knowledge, the reduction leads to an NP computation. Therefore, if applying embedding as backdoor attacks and extraction as defence, our results suggest a complexity gap (P vs. NP) between the attack and defence when working with tree ensemble classifiers. We apply our algorithms to a diverse set of datasets to validate our conclusion extensively.

Hidden Music in Early Elizabethan Tragedy

In modern times, scholars have widely regarded early Elizabethan tragedy, like Thomas Norton and Thomas Sackville’s Gorboduc (1561/62) and its successors at the Inns of Court, as verbose and unlyrical. Those criticisms may reflect an incomplete understanding of the original performance tradition, however. Like Senecan tragedies from this period, those plays include act-ending choruses, mostly in pentameter and in various stanza configurations. This study proposes that in the English tragedies, at least, those choruses were very likely sung, most probably to tunes from the emerging repertoire of metrical psalms. These findings would significantly affect the character of such plays and how they are perceived by scholars and audiences alike.

Assessment of Modified RELAP5 MOD3.2 Against LBE Natural Circulation Test Loops

Motivated by the significant natural circulation capability of lead–bismuth eutectic (LBE)–cooled systems, the RELAP5 MOD3.2 code was modified for the analysis of LBE-cooled reactors and non-nuclear systems. The thermo-physical properties of LBE have been incorporated into the code without affecting the code’s original performance; new heat transfer correlations for liquid metal have been implemented. For the purpose of validating the modified code, experimental results of two different LBE natural circulation test loops were compared with the code simulation results. The first one was a natural circulation setup process test at a power of 22.5 kW performed at the Natural Circulation Experimental (NACIE) facility. The simulated inlet and outlet LBE temperatures across the heat source and mass flow rate of LBE agreed well with the test data. The second one was natural circulation conditions under five different power levels conducted at the Natural Circulation Capability Loop (NCCL) facility. The LBE temperature difference and mass flow rate under different power levels predicted by the code were consistent with the experimental data. Generally speaking, the modified code gives acceptable results, and the code could be applied for further LBE systems thermal-hydraulic analysis.

“Kox Kwai Kauv Kox Kwai “:1 Ecopoetic Symbolisation in Pgaz K’nyau Oral Poetry

This article* presents the transcription, translation, and annotation of an original performance of hta, a traditional form of oral poetry in Sgaw, the language of the Pgaz K’Nyau (Karen) people of northern Thailand. This performance was recorded during ethnopoetic fieldwork carried out in two villages in the province of Chiang Rai.2 The hta is then analysed to understand the operations of ecopoetic symbolisation that bring particular nonhumans into the domain of human language. This analysis reveals that a metaphorical mode of symbolisation is extensively used throughout the hta to overcome human/nonhuman allotopies by means of implicit or explicit semic transformations. This seems to indicate that a naturalistic mode of identification underlies the whole poem, a conclusion that calls into question the essentialising and mythifying portrayal of the Pgaz K’Nyau as pre-modern and animistic indigenous stewards.

Optimal 1-NN prototypes for pathological geometries

Using prototype methods to reduce the size of training datasets can drastically reduce the computational cost of classification with instance-based learning algorithms like the k-Nearest Neighbour classifier. The number and distribution of prototypes required for the classifier to match its original performance is intimately related to the geometry of the training data. As a result, it is often difficult to find the optimal prototypes for a given dataset, and heuristic algorithms are used instead. However, we consider a particularly challenging setting where commonly used heuristic algorithms fail to find suitable prototypes and show that the optimal number of prototypes can instead be found analytically. We also propose an algorithm for finding nearly-optimal prototypes in this setting, and use it to empirically validate the theoretical results. Finally, we show that a parametric prototype generation method that normally cannot solve this pathological setting can actually find optimal prototypes when combined with the results of our theoretical analysis.

Structural Reinforcement Effect of a Flexible Strain Sensor Integrated with Pneumatic Balloon Actuators for Soft Microrobot Fingers

Motion capture of a robot and tactile sensing for a robot require sensors. Strain sensors are used to detect bending deformation of the robot finger and to sense the force from an object. It is important to introduce sensors in effective combination with actuators without affecting the original performance of the robot. We are interested in the improvement of flexible strain sensors integrated into soft microrobot fingers using a pneumatic balloon actuator (PBA). A strain sensor using a microchannel filled with liquid metal was developed for soft PBAs by considering the compatibility of sensors and actuators. Inflatable deformation generated by PBAs, however, was found to affect sensor characteristics. This paper presents structural reinforcement of a liquid metal-based sensor to solve this problem. Parylene C film was deposited into a microchannel to reinforce its structure against the inflatable deformation caused by a PBA. Parylene C deposition into a microchannel suppressed the interference of inflatable deformation. The proposed method enables the effective combination of soft PBAs and a flexible liquid metal strain sensor for use in microrobot fingers.

Optimization Approach for Solvent Extraction Process of Oily Contaminated Soil with Addition of Biosurfactant

Objectives : Solvent extraction is a process in which not only enable to reduce oil contaminant from soil residue, but also capable to recover oil from soil matrix of oily contaminated soil which has opportunity to be reutilized. Optimization process has been simulated by previous studies related to type and dosage of solvents, variances of temperature, additional of surfactants, and other related parameters to increase oil removal from oily contaminated soil. This study seeks an approach of optimization for solvent extraction process to oily contaminated solid waste by conducting statistical analysis into laboratory experimentation from perspective of Total Petroleum Hydrocarbon (TPH) removal.Method : Biosurfactant became single extractors for multistage extraction process and also combined with other solvents which are acetone and toluene. Mixing method that utilized during the study was combination between horizontal shaking at 150 rpm in 15 min duration and centrifugation force at 1,570 g in 10 min duration. Statistical analysis were conducted to seek its multiple regression.Result : Study describing biosurfactant performance single extractor by using multistage extraction process achieve 77% TPH removal, while combination of biosurfactant and solvent extraction by using toluene and acetone also capable to increase TPH removal 7% higher from original performance of both toluene and acetone at solvent extraction.Conclusion : Surfactant and solvents combination is promising to improve TPH removal, while statistics analysis that implemented to observed extraction process has possibility to be used for engineering higher efficiency of extraction process.

Individual and Communal Freedom and the Performance History of the St. Matthew Passion by Bach and Mendelssohn

When Felix Mendelssohn Bartholdy performed Johann Sebastian Bach’s St. Matthew Passion in the concert hall of the Berlin Singakademie in 1829, he not only transferred a piece of liturgical music into a secular space, but he also made numerous cuts that changed the theological profile of Bach’s composition. The essay explores the theology of the St. Matthew Passion in the context of early eighteenth-century theology and gives an overview of the original performance conditions and the audiences at the performances in Bach’s time. The second half of the essay analyses how these parameters changed when Mendelssohn conducted the Passion in 1829. It becomes clear that the sociological profile of the audience (educated middle and upper class who had to pay money to attend the performance) remained essentially the same, while the theology shifted from a focus on the freedom of the individual in Bach’s time to an emphasis on the community (congregation, Volk, nation) in the adapted version the Singakademie presented to its listeners in 1829.

Can We Predict Armed Conflict? How the First 9 Years of Published Forecasts Stand Up to Reality

Can we predict civil war? This article sheds light on this question by evaluating 9 years of, at the time, future predictions made by Hegre et al. (2013) in 2011. We evaluate the ability of this study to predict observed conflicts in the 2010–2018 period, using multiple metrics. We also evaluate the original performance evaluation, i.e., whether the performance measures presented by Hegre et al. hold in this new 9-year window. Overall, we conclude that Hegre et al. were able to produce meaningful and reasonably accurate predictions of armed conflict. Of course, they did not always hit the mark. We find that the model has performed worse in predicting low level incidence of conflict than in predicting major armed conflict. The model also failed to predict some important broader regional shifts. These, however, represent important insights for future research and illustrate the utility in predictive models for both testing and developing theory.

The audiovisual recreation of operas filmed in theaters: An analysis of Don Giovanni by W. A. Mozart

This article analyzes select opera films produced during the 20th and 21st centuries by international opera houses in order to determine whether, when the diegesis of these films is recreated, there is also an attempt made for the films to reflect the conventions of operatic performance. We performed content analysis of the end of the first act in 29 filmed versions of the opera Don Giovanni by W.A. Mozart by evaluating 44 variables related to three categories that are central to translating the original story to the new audiovisual discourse: recreation of time and space and use of scenery. The main results reveal that in translating Don Giovanni to audiovisual media, the films continue to be influenced by the institutional conventions of operatic theater. In relying on the original performance, these films, even in the 21st century, do not exhibit full narrative autonomy. Our article proposes several ways to better adapt—from a cinematographic perspective—these scenic representations based on elements such as depth of field or scenery, without the need for greater resources than those already available to opera houses.