Estimating Degradation of Machine Learning Data Assets

Large-scale adoption of Artificial Intelligence and Machine Learning (AI-ML) models fed by heterogeneous, possibly untrustworthy data sources has spurred interest in estimating degradation of such models due to spurious, adversarial, or low-quality data assets. We propose a quantitative estimate of the severity of classifiers’ training set degradation: an index expressing the deformation of the convex hulls of the classes computed on a held-out dataset generated via an unsupervised technique. We show that our index is computationally light, can be calculated incrementally and complements well existing ML data assets’ quality measures. As an experimentation, we present the computation of our index on a benchmark convolutional image classifier.

Studying Up Machine Learning Data

Research in machine learning (ML) has argued that models trained on incomplete or biased datasets can lead to discriminatory outputs. In this commentary, we propose moving the research focus beyond bias-oriented framings by adopting a power-aware perspective to "study up" ML datasets. This means accounting for historical inequities, labor conditions, and epistemological standpoints inscribed in data. We draw on HCI and CSCW work to support our argument, critically analyze previous research, and point at two co-existing lines of work within our research community \,---\,one bias-centered, the other power-aware. We highlight the need for dialogue and cooperation in three areas: data quality, data work, and data documentation. In the first area, we argue that reducing societal problems to "bias" misses the context-based nature of data. In the second one, we highlight the corporate forces and market imperatives involved in the labor of data workers that subsequently shape ML datasets. Finally, we propose expanding current transparency-oriented efforts in dataset documentation to reflect the social contexts of data design and production.

Analysis of the Use of E-Learning as a Historical Learning Medium during the Covid-19 Pandemic

The purpose of this study was to see how E-learning applications were used in history lectures during the Covid-19 period. This research uses descriptive-evaluative research strategy with Mixed Methods technique (quantitative and qualitative). The descriptive-evaluative analysis in this study is limited to the E-Learning program as a historical learning medium. The subjects of this study were teachers of history subjects and students at SMA Negeri 1 Kota Padang who used E-Learning. Data was collected through interviews, documentation, and distributing questionnaires. This study uses descriptive statistical analysis, namely calculating the amount of data obtained from questionnaire data and then evaluating the data in the form of percentages. The influence of online learning with E-Learning applications on students’ historical thinking skills and historical awareness abilities is substantial. Furthermore, using E-Learning technology into history lessons may aid students in developing a stronger sense of national identity. As a result, it is possible to infer that the E-learning application had a significant impact on history learning during the COVID-19 epidemic.

Generative Design of Stable Semiconductor Materials Using Deep Learning And DFT

Semiconductor device technology has exceptionally developed in complexity since discovering the bipolar transistor. With the rapid advancement of various technologies, semiconductors with distinct properties are essential. Recently, deep-learning, data-mining, and density functional theory (DFT)- based high-throughput calculations were widely performed to discover potential semiconductors for diverse applications. CubicGAN is a generative adversarial network where high-throughput analyses were done to uncover mechanically and dynamically stable materials with the assistance of DFT. In our work, we screened the semiconductors using a binary classifier from materials found from the CubicGAN. Next, we performed DFT computations to study their thermodynamic stability based on energy-above-hull and formation energy. According to our studies, 12 stable semiconductors were found with a particular class of materials, which we label as AA′MH6. Those are BaNaRhH6, BaSrZnH6, BaCsAlH6, SrTlIrH6, KNaNiH6, NaYRuH6, CsKSiH6, CaScMnH6, YZnMnH6, NaZrMnH6, AgZrMnH6, AgZrMnH6, and ScZnMnH6. It could be shown that AA′MH6 with M=Mn and NaYRuH6 semiconductors have considerably different structural, mechanical, and thermodynamic properties compared to the rest of the AA′MH6 semiconductors. In this study, The maximum bandgap found was approximately 3.3 eV from KNaNiH6, while the minimum bandgap was about 1.3 eV from CaScMnH6. BaNaRhH6, BaCsAlH6, CsKSiH6, KNaNiH6, and NaYRuH6 were identified as wide-bandgap semiconductors, where bandgaps are greater than 2 eV. Furthermore, BaSrZnH6 and KNaNiH6 are a direct bandgap semiconductors, whereas other AA′MH6 semiconductors exhibit indirect bandgaps.

Methodologies to Associate COVID-19 Spreading Data to Space and Scale

Since the onset of the COVID-19 pandemic, the correlation between the spread of the SARS-Cov-2 virus and a number of epidemiological parameters has been a key tool for understanding the dynamics of its flow. This information has assisted local authorities in making policy decisions for the containment of its expansion. Several methods have been used including topographical data, artificial intelligence and machine learning data, and epidemiological tools to analyze factors facilitating the spread of epidemic at a local and global scale. The aim of this study is to use a new tool to assess and categorize the incoming epidemiological data regarding the spread of the disease as per population densities, spatial and topographical morphologies, social and financial activities, population densities and mobility between regions. These data will be appraised as risk factors in the spread of the disease on a local and a global scale.

Digitalism and Jobs of the Future

There has been a strong relationship between digitalism and the future of jobs. Reports by OECD and WEF examined the jobs in the coming decades, and the findings show that there is a completely new order in the professions that we are not familiar with. In addition, how the impacts of artificial intelligence (AI), machine learning, data science, and robotics have affected labour, the market is analyzed. The findings in the reports clearly would affect the careers of the next generations. With the post-pandemic developments and the rapid advancement of technology in many areas worldwide, digitalization has gained significant momentum. This situation manifested itself in professions and workforce. However, it is obvious that in the coming years, with digitalization, many occupational groups and accordingly, differences in skills will be seen. While some occupational groups disappear completely, it is seen that some new occupational groups will emerge and, some will transform.

Stock Prediction Model Using TensorFlow

In Stock Market Prediction, the aim is to predict the future value of the financial stocks of a company. The recent trend available in market prediction technologies is that the use of machine learning that makes predictions on the basis of values of current stock exchange indices by training on their previous values. Machine learning itself employs completely different models to create prediction easier and authentic. The paper focuses on the use of Regression and LSTM based Machine learning to predict stock values. Considering the factors are open, close, low, high and volume. Keywords: Stock Prediction, Machine Learning, Data Visualization, Yahoo Finance Dataset

An Interview with Dr. Shipeng Yu, Winner of ACM SIGKDD 2021 Service Award

Shipeng Yu, Ph.D. is the recipient of the 2021 ACM SIGKDD Service Award, which is the highest service award in the field of knowledge discovery and data mining. Conferred annually on one individual or group in recognition of outstanding professional services and contributions to the field of knowledge discovery and data mining, Dr. Yu was honored for his years of service and many accomplishments as general chair of KDD 2017 and currently as sponsorship director for SIGKDD. Dr. Yu is Director of AI Engineering, Head of the Growth AI team at LinkedIn, the world's largest professional network. He sat down with SIGKDD Explorations to discuss how he first got involved in the KDD conference in 2006, the benefits and drawbacks of virtual conferences, his work at LinkedIn, and KDD's place in the field of machine learning, data science and artificial intelligence.