Five days after Russia launched its full-scale invasion of Ukraine, a year ago this week, US-based facial recognition company Clearview AI offered the Ukrainian government free access to its technology, suggesting that it could be used to reunite families, identify Russian operatives, and fight misinformation. Soon afterward, the Ukraine government revealed it was using the technology to scan the faces of dead Russian soldiers to identify their bodies and notify their families. By December 2022, Mykhailo Fedorov, Ukraine's vice prime minister and minister of digital transformation, was tweeting a picture of himself with Clearview AI's CEO Hoan Ton-That, thanking the company for its support. Accounting for the dead and letting families know the fate of their relatives is a human rights imperative written into international treaties, protocols, and laws like the Geneva Conventions and the International Committee of the Red Cross' (ICRC) Guiding Principles for Dignified Management of the Dead. It is also tied to much deeper obligations.

President Biden was asked by PBS' Judy Woodruff about his classified documents scandal in his first interview since the controversy broke last month. Members of Congress have taken a sharp tone towards the possible mishandling of classified documents by President Biden, former President Trump and former Vice President Pence in recent weeks, and many have called for legislative action to prevent such occurrences from happening again. It's unclear, however, what action Congress could take, considering the level of classification of documents has been managed by the executive branch of the U.S. government since the time of the late President Franklin Roosevelt, and no members speaking out on the issue have provided any specifics as to what can actually be done to address the problem. "Holding classified documents in anything other than a very secure setting is a risk to national security that is very serious and needs to be resolved. And there are a number of elements to that," Sen. Mitt Romney, R-Utah, said last month after reports surfaced that Pence was the latest to be in possession of classified documents, and just weeks after some had been found at Biden's home and former office. Biden, Pence, and Donald Trump are under intense scrutiny for classified documents being found at their personal properties.

Vietnam, a country in Southeast Asia, with an area of 311,699 square kilometres (120,348 square miles) has a population of over 97 million and if you ever visit this beautiful country,you will soon appreciate that there are also over 65 million registered motorbikes, with many families having a motorbike for each family member. This article summarizes a number of research sources to give a sense of where Vietnam's leadership is in the field of AI – and all indications are its status is evolving and blossoming. Where is Vietnam in terms of Information Communication and Technology (ICT) companies, and how are they evolving their AI leadership position? According to the Ministry of Information and Communications, the revenue of the ICT industry in 2021 was $136,153 million USD, a solid increase compared to $124,678 million USD in 2020. It is also estimated that the ratio of Vietnam's value in ICT revenue reached 24.65%, a significant increase compared to previous years.

Laura Ingraham rips into the Democrats push for'equity' and what they actually mean by the term on'The Ingraham Angle.' President Biden signed an executive order on Thursday that critics warn will allow for the creation of woke artificial intelligence (AI) that "promotes racial division and discrimination." The order directs all federal agencies to establish a yearly "equity action plan" aimed at helping "underserved communities." In one section under "Embedding Equity into Government-wide Processes," the Director of the Office of Management and Budget is instructed "to support equitable decision-making, promote equitable deployment of financial and technical assistance, and assist agencies in advancing equity, as appropriate and wherever possible." The section gives further instructions regarding AI, "When designing, developing, acquiring, and using artificial intelligence and automated systems in the Federal Government, agencies shall do so, consistent with applicable law, in a manner that advances equity." President Joe Biden signed the executive order "Further Advancing Racial Equity and Support for Underserved Communities Through the Federal Government," February 16, 2023.

Pre-trained multilingual language models (e.g., mBERT, XLM-RoBERTa) have significantly advanced the state-of-the-art for zero-shot cross-lingual information extraction. These language models ubiquitously rely on word segmentation techniques that break a word into smaller constituent subwords. Therefore, all word labeling tasks (e.g. named entity recognition, event detection, etc.), necessitate a pooling strategy that takes the subword representations as input and outputs a representation for the entire word. Taking the task of cross-lingual event detection as a motivating example, we show that the choice of pooling strategy can have a significant impact on the target language performance. For example, the performance varies by up to 16 absolute $f_{1}$ points depending on the pooling strategy when training in English and testing in Arabic on the ACE task. We carry out our analysis with five different pooling strategies across nine languages in diverse multi-lingual datasets. Across configurations, we find that the canonical strategy of taking just the first subword to represent the entire word is usually sub-optimal. On the other hand, we show that attention pooling is robust to language and dataset variations by being either the best or close to the optimal strategy. For reproducibility, we make our code available at https://github.com/isi-boston/ed-pooling.

To date, much of the algorithmic fairness literature has focused on the fairness of classification systems which are used, for example, to decide whether a person should be granted a loan or be released from prison on bail. Even in cases where such classification decisions are based on risk score models - such as in the highly influential COMPAS case [5, 11, 16] - their fairness is typically considered a function of the decisions, or classifications, made by the system. Of course, any risk score model can be turned into a classifier by selecting a probability threshold (in binary classification) or predicting the most likely outcome (in multi-class classification). Nevertheless, we argue here that it is worthwhile to distinguish between these two settings and consider the fairness of risk models independent of their downstream use, be it as the basis for a classifier or otherwise. We discuss notions of fairness for risk scores as well as their relationship to classical, classification-level notions of fairness, and we develop robust tools to empirically quantify risk score fairness. We illustrate our methodology in two case studies, one situated in the criminal justice system and one in healthcare. Why distinguish between fair models and fair decisions? In the statistical literature, it is generally considered desirable to distinguish between inference (e.g., identifying a risk score model) and subsequent decision-making (e.g., deriving a classification from a risk score model): while the former represents a purely statistical task, the latter depends on subjective

Data scarcity is a problem that occurs in languages and tasks where we do not have large amounts of labeled data but want to use state-of-the-art models. Such models are often deep learning models that require a significant amount of data to train. Acquiring data for various machine learning problems is accompanied by high labeling costs. Data augmentation is a low-cost approach for tackling data scarcity. This paper gives an overview of current state-of-the-art data augmentation methods used for natural language processing, with an emphasis on methods for neural and transformer-based models. Furthermore, it discusses the practical challenges of data augmentation, possible mitigations, and directions for future research.

The development of annotated datasets over the 21st century has helped us truly realize the power of deep learning. Most of the datasets created for the named-entity-recognition (NER) task are not domain specific. Finance domain presents specific challenges to the NER task and a domain specific dataset would help push the boundaries of finance research. In our work, we develop the first high-quality NER dataset for the finance domain. To set the benchmark for the dataset, we develop and test a weak-supervision-based framework for the NER task. We extend the current weak-supervision framework to make it employable for span-level classification. Our weak-ner framework and the dataset are publicly available on GitHub and Hugging Face.

Machine Learning models face increased concerns regarding the storage of personal user data and adverse impacts of corrupted data like backdoors or systematic bias. Machine Unlearning can address these by allowing post-hoc deletion of affected training data from a learned model. Achieving this task exactly is computationally expensive; consequently, recent works have proposed inexact unlearning algorithms to solve this approximately as well as evaluation methods to test the effectiveness of these algorithms. In this work, we first outline some necessary criteria for evaluation methods and show no existing evaluation satisfies them all. Then, we design a stronger black-box evaluation method called the Interclass Confusion (IC) test which adversarially manipulates data during training to detect the insufficiency of unlearning procedures. We also propose two analytically motivated baseline methods~(EU-k and CF-k) which outperform several popular inexact unlearning methods. Overall, we demonstrate how adversarial evaluation strategies can help in analyzing various unlearning phenomena which can guide the development of stronger unlearning algorithms.

An increasing number of data science and machine learning problems rely on computation with tensors, which better capture the multi-way relationships and interactions of data than matrices. When tapping into this critical advantage, a key challenge is to develop computationally efficient and provably correct algorithms for extracting useful information from tensor data that are simultaneously robust to corruptions and ill-conditioning. This paper tackles tensor robust principal component analysis (RPCA), which aims to recover a low-rank tensor from its observations contaminated by sparse corruptions, under the Tucker decomposition. To minimize the computation and memory footprints, we propose to directly recover the low-dimensional tensor factors -- starting from a tailored spectral initialization -- via scaled gradient descent (ScaledGD), coupled with an iteration-varying thresholding operation to adaptively remove the impact of corruptions. Theoretically, we establish that the proposed algorithm converges linearly to the true low-rank tensor at a constant rate that is independent with its condition number, as long as the level of corruptions is not too large. Empirically, we demonstrate that the proposed algorithm achieves better and more scalable performance than state-of-the-art matrix and tensor RPCA algorithms through synthetic experiments and real-world applications.