There is an amazing article in the NY Times today about how the US government helped cover-up the intimate relationship between the Colombian paramilitaries and the Uribe government. If you read the article (along with this article) its clear that Uribe had a bunch of paramilitary leaders extradited to the US (in the middle of the night without the Colombian supreme court’s approval) because the paramilitary leaders essentially implicating the government and large fractions of the Colombian elite. A few amazing extracts…
Much to Mr. Uribe’s consternation, the leaders then began confessing not only their war crimes but also their ties to his allies and relatives. The Colombian Supreme Court — responsible for investigating lawmakers — undertook an aggressive “para-politics” inquiry that ensnared many in the president’s coalition…
They were, collectively, going to deliver testimony that directly implicated Uribe,” said Senator Iván Cepeda, who founded the influential Movement for Victims of State Crimes. “Afterward, the authorities entered the cells where they had their computers, their USBs, and they took it all. All the work they had been doing, all the proof they were going to present to Justice, disappeared.
“Those extraditions marked a before and an after,” Mr. Cepeda continued. “If the intention really was to achieve silence and impunity, it was obtained to a high degree. Only now, after so many years, are we starting to see results.”…
And the US involvement and facilitation is amazing as well.
After years of declining to turn over the men, President Uribe had suddenly made an urgent appeal to the Americans: Those wanted paramilitary leaders? Take them. Immediately.
He wanted them picked up after the Colombian Supreme Court closed for the day and flown out before the court resumed the next morning, said a United States official who agreed to speak only on the condition of anonymity. Mr. Uribe said he feared that the court might block the extraditions if they did not act hurriedly.
The Americans snapped into action, though it was a logistical challenge. They needed to “move the men from four corners of Colombia to one location, then the aircraft needed to go wheels up with all prisoners onboard prior to the court opening for business the next day,” the official said.
At one point, the official said, the Drug Enforcement Administration had six aircraft “in play” to ferry the men through Bogotá and Guantánamo, Cuba, to courthouses in Florida, New York, Texas and Washington, D.C.
You have to read the rest of the story to really realize what the Plan Colombia and the war on drugs was really about — protecting the elite and their wealth.
This semester I am teaching a new class, Machine Learning (ML) for Physicists. The website for the class is here. The class is a response to students requests and desires to learn these skills.
The reality is that most physics graduate students end up outside academia, and the tenure-track professorship is the exception not the rule. For this reason, its important to impart useful skills to our graduate students. This class is my contribution to that goal. To my delight, the graduate students at BU have responded and its the most popular graduate physics course this semester (its actually completely full)!
Of course, ML is also useful in physics. Here at BU, we are doing a number of really exciting things combining ML and physics that I will talk more about in later blog posts. We are particularly excited about new work we are doing using Reinforcement Learning to think about quantum dynamics.
The class is my attempt at a short synthesis that builds on the significant base of knowledge that a typical graduate student already has in Statistical Physics. One of the things that struck me is that most books on the subject tend to be long +700 pages and inevitably end up being a comprehensive overview off methods and techniques. Another observation is that books tend to basically fall into two camps: the Bayesian camp and the statistical learning/Frequentist camp. Very few books or reviews treat both these (useful) perspective on a equal footing.
I hope my class can overcome these limitations. To do so, we are going to take the physicist approach of solving a few, well-chosen and informative canonical examples (linear regression, logistic regression, clustering) over and over again from different perspectives, using different techniques. After all, how many ways can a physicist solve the Ising model or the harmonic oscillator?
Another thing we will be doing is “hands-on” exercises using Python Notebooks. I just finished making my first notebook, “Why Prediction is Difficult?”. You can find it here as html or download it here. The point of the notebook is to force students to think about why fitting existing data well is not the same thing as making good predictions.
Anyway, I will try to post updates about the class on this blog as well as useful concepts. I hope to turn this class into a useful 80 page review or so. If you have suggestions, please let me know by email or comment.
Lets see if I can regularly update this thing. Anyway, here are two recent articles I have written, an obituary for the late great Richard Levins – The People’s Scientist. He is everything I aspire to be as a scientist. I hope I can live up to his example.
I also have a review of Mathbabe‘s excellent new book, Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy coming out in the In These Times.
My new article on the Science and politics of big data is out at Jacobin. Check it out here and tell me what you think.
Quanta Magazine just wrote a piece on our arXiv preprint relating Deep Learning and the Renormalization group. To find out more, please check out the piece or our preprint. Deep learning techniques have recently yielded record-breaking results on a diverse set of difficult machine learning tasks including computer vision, speech recognition, and natural language processing. Deep learning is one of the most exciting new techniques to emerge for unsupervised learning and companies such as Google, Microsoft, and Facebook have invested heavily in this field. It is commonly touted in both the popular press and as well as in the scientific community as a major breakthrough for machine learning.
Despite the enormous success of deep learning, relatively little is understood theoretically about why it is so successful at uncovering relevant features in structured data. In our preprint, we show that deep learning is intimately related to one of the most important and successful techniques in theoretical physics, the Renormalization Group (RG). This suggests that deep learning architectures may be employing a generalized RG-like scheme to learn relevant features from data! For more, check out the Quanta article and our preprint.
The New Yorker recently penned an extremely misleading piece painting anti-GMO activists as anti-science hacks. It focuses on the prominent anti-GMO activist Vandana Shiva. It regurgitates hackneyed arguments from the Biotech industry painting people against GMOs are irrational Luddites who are harming the world. Luckily, Vandana Shiva has written a spirited and extremely devastating response. Everyone should read the response.
While I don’t fully agree with all of Vandana Shiva’s politics, between Monsanto and Shiva, I think there is no choice at all: one is an amoral corporation who ruthlessly prioritizes profits over people, the other a dedicated activist who has helped amplify the voices of everyday peasants devastated by neoliberal science and development policies. Give me Vandana Shiva everytime!