Collaborating Authors


HealthITAnalytics: Machine Learning Can Streamline PTSD Diagnosis in Veterans - FedHealthIT


"Using machine learning, researchers were able to cut six of the 20 questions used to diagnose post-traumatic stress disorder (PTSD) while still maintaining accuracy in a veteran population, according to a study published in Assessment." "PTSD impacts eight million adults in the US, the researchers stated, including hundreds of thousands of veterans of the conflicts in Iraq and Afghanistan. With the onset of the COVID-19 pandemic, PTSD symptoms are also increasing among the general population, the team noted." "However, diagnosing PTSD is time-consuming – the process typically takes 30 minutes or more, which is too long for most clinical visits." "To streamline PTSD diagnosis, researchers from the VA Boston Healthcare System and the Boston University School of Public Health (BUSPH) set out to develop a machine learning tool that would make the process more efficient." "The team used data from the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, fifth edition (SCID-5) assessments of 1,265 veterans who served in Iraq and Afghanistan."

All-Girl Robotics Team In Afghanistan Works On Low-Cost Ventilator ... With Car Parts

NPR Technology

Elham Mansoori, member of Afghan Dreamers, an all-girls robotics team in Afghanistan, works on their prototype of a ventilator. In Afghanistan, a group of teenage girls are trying to build a mechanized, hand-operated ventilator for coronavirus patients, using a design from M.I.T. and parts from old Toyota Corollas. It sounds like an impossible dream, but then again, the all-girls robotics team in question is called the "Afghan Dreamers." Living a country where two-thirds of adolescent girls cannot read or write, they're used to overcoming challenges. The team of some dozen girls aged 15 to 17 was formed three years ago by Roya Mahboob, an Afghan tech entrepreneur who heads the Digital Citizen Fund, a group that runs classes for girls in STEM and robotics and oversees and funds the Afghan Dreamers.

Ventilators from old car parts? Afghan girls pursue prototype amid coronavirus lockdown

The Japan Times

Kabul – On most mornings, Somaya Farooqi and four other teenage girls pile into her dad's car and head to a mechanic's workshop. They use back roads to skirt police checkpoints set up to enforce a lockdown in their city of Herat, one of Afghanistan's hot spots of the coronavirus pandemic. The members of Afghanistan's prize-winning girls' robotics team say they're on a life-saving mission -- to build a ventilator from used car parts and help their war-stricken country battle the virus. "If we even save one life with our device, we will be proud," said Farooqi, 17. Their pursuit of a low-cost breathing machine is particularly remarkable in conservative Afghanistan.

The Bethe and Sinkhorn Permanents of Low Rank Matrices and Implications for Profile Maximum Likelihood Machine Learning

In this paper we consider the problem of computing the likelihood of the profile of a discrete distribution, i.e., the probability of observing the multiset of element frequencies, and computing a profile maximum likelihood (PML) distribution, i.e., a distribution with the maximum profile likelihood. For each problem we provide polynomial time algorithms that given $n$ i.i.d.\ samples from a discrete distribution, achieve an approximation factor of $\exp\left(-O(\sqrt{n} \log n) \right)$, improving upon the previous best-known bound achievable in polynomial time of $\exp(-O(n^{2/3} \log n))$ (Charikar, Shiragur and Sidford, 2019). Through the work of Acharya, Das, Orlitsky and Suresh (2016), this implies a polynomial time universal estimator for symmetric properties of discrete distributions in a broader range of error parameter. We achieve these results by providing new bounds on the quality of approximation of the Bethe and Sinkhorn permanents (Vontobel, 2012 and 2014). We show that each of these are $\exp(O(k \log(N/k)))$ approximations to the permanent of $N \times N$ matrices with non-negative rank at most $k$, improving upon the previous known bounds of $\exp(O(N))$. To obtain our results on PML, we exploit the fact that the PML objective is proportional to the permanent of a certain Vandermonde matrix with $\sqrt{n}$ distinct columns, i.e. with non-negative rank at most $\sqrt{n}$. As a by-product of our work we establish a surprising connection between the convex relaxation in prior work (CSS19) and the well-studied Bethe and Sinkhorn approximations.

Space and Time Efficient Kernel Density Estimation in High Dimensions

Neural Information Processing Systems

Recently, Charikar and Siminelakis (2017) presented a framework for kernel density estimation in provably sublinear query time, for kernels that possess a certain hashing-based property. However, their data structure requires a significantly increased super-linear storage space, as well as super-linear preprocessing time. These limitations inhibit the practical applicability of their approach on large datasets. In this work, we present an improvement to their framework that retains the same query time, while requiring only linear space and linear preprocessing time. We instantiate our framework with the Laplacian and Exponential kernels, two popular kernels which possess the aforementioned property.

What do we look for in a 'good' robot colleague?


With a tank-like continuous track and an angular arm reminiscent of the Pixar lamp, the lightweight PackBot robot was designed to seek out, defuse and dispose of the improvised explosive devices, or IEDs, that killed and injured thousands of coalition soldiers during the wars in Iraq and Afghanistan. Bomb disposal was and is highly dangerous work, but the robot could take on the riskiest parts while its human team controlled it remotely from a safer distance. US Army explosive ordinance disposal technician Phillip Herndon was assigned a PackBot during his first tour in Iraq. Herndon's team named their robot Duncan, after a mission when the robot glitched and began spinning in circles, or doughnuts (doughnuts led to Dunkin Donuts, hence Duncan). His fellow bomb disposal techs named theirs too, and snapped photos of themselves next to robots holding Xbox controllers, dressed in improvised costumes or posing with a drink in their claws.

AI apocalypse: Ex-Google worker fears 'killer robots' could cause 'mass atrocities'


"Although I was not directly involved in speeding up the video footage recognition I realised that I was still part of the kill chain; that this would ultimately lead to more people being targeted and killed by the US military in places like Afghanistan." The former Google engineer predicts autonomous weapons currently in development pose a far greater risk to humanity than remote-controlled drones. She outlined how external forces ranging from changing weather systems to machines being unable to work out complex human behaviour might throw killer robots off course, with potentially fatal consequences. She told The Guardian: "You could have a scenario where autonomous weapons that have been sent out to do a job confront unexpected radar signals in an area they are searching; there could be weather that was not factored into its software or they come across a group of armed men who appear to be insurgent enemies but in fact are out with guns hunting for food. "The machine doesn't have the discernment or common sense that the human touch has.

Prediction of adverse events in Afghanistan: regression analysis of time series data grouped not by geographic dependencies Machine Learning

The aim of this study was to approach a difficult regression task on highly unbalanced data regarding active theater of war in Afghanistan. Our focus was set on predicting the negative events number without distinguishing precise nature of the events given historical data on investment and negative events per each of predefined 400 Afghanistan districts. In contrast with previous research on the matter, we propose an approach to analysis of time series data that benefits from non-conventional aggregation of these territorial entities. By carrying out initial exploratory data analysis we demonstrate that dividing data according to our proposal allows to identify strong trend and seasonal components in the selected target variable. Utilizing this approach we also tried to estimate which data regarding investments is most important for prediction performance. Based on our exploratory analysis and previous research we prepared 5 sets of independent variables that were fed to 3 machine learning regression models. The results expressed by mean absolute and mean square errors indicate that leveraging historical data regarding target variable allows for reasonable performance, however unfortunately other proposed independent variables does not seem to improve prediction quality.

List-Decodable Subspace Recovery via Sum-of-Squares Machine Learning

An influential recent line of work [KLS09, ABL13, DKK 16, LRV16, CSV17, KS17a, KS17b, HL17, DKK 18, DKS17a, KKM18] has focused on designing algorithms for basic statistical estimation tasks in the presence of adversarial outliers. This has led to a body of work on outlier-robust estimation of basic parameters of distributions such as mean, covariance [DKK 16, DKS17b, CDG19, DKK 17, DKK 18, CDGW19] and moment tensors [KS17b] along with applications to "downstream" learning tasks such as linear and polynomial regression [DKS17c, KKM18, DKK 19, PSBR18]. The upshot of this line of work is a detailed understanding of efficient robust estimation when the fraction of inliers ( 1/2), but a fixed fraction of arbitrary adversarial outliers in the input data. In this work, we focus on the harsher list-decodable estimation model where the fraction of inliersαis 1/2 - i.e.,a majority of the input sample are outliers. First considered in [BBV08] in the context of clustering, this was proposed as a model for untrusted data in a recent influential recent work of Charikar, Steinhardt and Valiant [CSV17]. Since unique recovery is informationtheoretically impossible in this setting, the goal is to recover a small (ideally O(1/α)) size list of parameters one of which is guaranteed to be close to those of the inlier distribution. A recent series of works have resulted in a high-level blueprint based on the sum-of-squares method for listdecodable estimation yielding algorithms for list-decodable mean estimation [DKS18, KS17a] and linear regression [KKK19, RY20]. We extend this line of work by giving the first efficient algorithm for list-decodable subspace recovery. In this setting, we are given data with α fraction inliers generated i.i.d.

Military artificial intelligence can be easily and dangerously fooled


Kanaan is generally very bullish about AI, partly because he knows firsthand how useful it stands to be for troops. Six years ago, as an Air Force intelligence officer in Afghanistan, he was responsible for deploying a new kind of intelligence-gathering tool: a hyperspectral imager. The instrument can spot objects that are normally hidden from view, like tanks draped in camouflage or emissions from an improvised bomb-making factory. Kanaan says the system helped US troops remove many thousands of pounds of explosives from the battlefield. Even so, it was often impractical for analysts to process the vast amounts of data collected by the imager.