“What the hell are we going to use lasers for except to kill people?” Jones said. “But scientists get cut the slack.”

I’m flabbergasted that someone who works on philosophy applied to a technological field, namely biomedical ethics, believes that the only use of lasers is to kill people. Perhaps she thinks that using lasers in surgery is unethical. Or, more likely, she is unaware of how basic research in science is actually funded in this country.

Certainly, there’s been a definite shift over time in how defense-related agencies have targeted their funds — they fund much less basic research (or basic applied research) and have focused more on deliverables and technologies that more directly support combat, future warriors, and the like. This presents important ethical questions for researchers who may oppose the use of military force (or how it has been used recently) but who are interested in problems that could be “spun” towards satisfying these new objectives from DARPA, ARO, ONR, and AFOSR. Likewise, there are difficult questions about the line between independent research and consulting work for companies who may fund your graduate students. Drawing sharp distinctions in these situations is hard — everybody has their own comfort zone.

Jones wrote an article on “Dirty Money” that tries to develop rules for when money is tainted and when it is not. She comes up with a checklist at the end of the article that says funds should not be accepted if they

1- are illegal or that operate illegally in one’s country, or when the funding violates a generally accepted doctrine signed by one’s country (keeping in mind there is sometimes a distinction between legally acceptable and morally acceptable); or

2- originate from a donor who adds controls that would conflict with the explicit or implicit goals of the project to be funded or that would conflict with the proper functioning of the project or the profession’s ethical guidelines.

This, she says, is “the moral minimum.” This framing (and the problem in general of funding centers) that she addresses sidesteps the ethical questions around research that is funded by writing proposals, and indeed the question of soliciting funds. Even in the world of charitable giving, the idea that funders wander through the desert with bags of money searching for fundees seems odd. I think the more difficult ethical quandary is that of *solicitation*. At a “moral minimum” the fundee has to think about these questions, but I think point 2 needs a lot more unpacking because of the chicken-and-egg question of matching proposed research to program goals.

I don’t want to sound so super-negative! I think it’s great that someone is looking at the ethics of the economics of how we fund research. It’s just that there’s a whole murkier lake beyond the murky pond of funding centers, and the moral issues of science/engineering funding are not nearly as simple as Jones’s remark indicates.

Filed under: Uncategorized ]]>

The piece talks about work in Agnieszka Roginska‘s lab at NYU, and in particular work from a paper from last year on measuring radiation patterns in piano soundboards. The radiation patterns are pretty but a bit hard to interpret, largely because I’m way out of the acoustical signal processing world. However, what’s interesting to me is that we’re still largely focused on overtones/cepstral coefficients. I wonder about how one might discover more interesting features to characterize this data. (I know someone will suggest deep learning but color me a little skeptical).

As a side note, one of the recent popular articles from JASA is on the acoustics of coffee roasting.

Filed under: Uncategorized ]]>

Aaron Roth and Cynthia Dwork’s Foundation and Trends monograph on differential privacy is now available.

Speaking of differential privacy, Shiva Kasiviswanathan and Adam Smith have a paper in the Journal of Privacy and Confidentiality on Bayesian interpretations of differential privacy risk.

Deborah Mayo has a post up on whether p-values are error probabilities.

Raymond Yeung is offering a Coursera course on information theory (via the IT Society).

A CS Theory take on Fano’s inequality from Suresh over at the GeomBlog.

Filed under: Uncategorized ]]>

We’re in the process of collecting reading materials — magazine articles, book chapters, blog posts, etc. for the students to read. We explicitly didn’t want it to be for “technical” students only. Do any readers of the blog have great articles suitable for first-year undergrads across all majors?

As the class progresses I will post materials here, as well as some snapshot of the discussion. It’s my first time teaching a class of this type (or indeed any undergraduates at Rutgers) so I’m excited (and perhaps a bit nervous).

On a side note, Edwin Starr’s shirt is awesome and I want one.

Filed under: Uncategorized ]]>

On a less grim note, the site’s promise to make your research “more visible” sounds a bit like SEO spam. Given the existence of Google Scholar, which is run by the SE that one would like to O, it seems slightly implausible.

Any readers want to weigh in on whether ResearchGate has been useful to them? Or is this mostly for people who don’t know how to make their own homepage with their papers on it (which is probably most faculty).

Filed under: Uncategorized ]]>

**Luka and the Fire of Life** [Salman Rushdie]. A re-read for me, this didn’t hold up as well the second time around. I much prefer *Haroun and the Sea of Stories*, which I can read over and over again.

**Boxers and Saints** [Gene Luen Yang]. A great two-part graphic novel about the Boxer Rebellion in China. Chances are you don’t know much about this history. You won’t necessarily get a history lesson from this book, but you will want to learn more about it.

**The Adventures of Augie March** [Saul Bellow]. After leaving Chicago I have decided to read more books set in Chicago so that I can miss it more. I had read this book before but it was a rushed job. This time I let myself longer a bit more over Bellow’s language. It’s epic and scope and gave me a view of Chicago and the Great Depression that I hadn’t had before. Indeed, given our current economic woes, it was an interesting comparison to see the similarities (the rich are still pretty rich, and if you can get employed by them, you may do ok) and the dissimilarities.

**The Idea Factory: Bell Labs and the Great Age of American Innovation** [John Gertner]. A history of Bell Labs and a must-read for researchers who work on anything related to computing, communications, or applied physics and chemistry. It’s not all rah-rah, and while Gertner takes the “profiles of the personalities” approaches to writing about the place, I am sure there will be things in there that would surprise even the die-hard Shannonistas who may read this blog…

Filed under: Uncategorized ]]>

**Communication and Compression Via Sparse Linear Regression**

*Ramji Venkataramanan*

This was on building codewords and codebooks out of a lower-complexity code dictionary where each codeword is a superposition of columns, one each from groups of size . Thus encoding is where is a sparse vector. I saw a talk by Barron and Joseph from a previous ISIT about this, but the framework extends to rate distortion (achieving the rate distortion function), and channel coding. The main point is to lower the complexity of the code at the expense of the gap to optimal rate — encoding and decoding are polynomial time but the rate gap for rate-distortion goes to zero as . Ramji gave a really nice and clear talk on this — I hope he puts the slides up!

**An Optimal Varentropy Bound for Log-Concave Distributions**

*Mokshay Madiman; Liyao Wang*

Mokshay’s talk was also really clear and excellent. For a distribution on , we can define . The entropy is the expectation of this random variable, and the varentropy is the variance. Their main result is a upper bound on the varentropu of log concave distributions . To wit, . This bound doesn’t depend on the distribution and is sharp if is a product of exponentials. They then use this to prove a universal bound on the deviation of from its expectation, which gives a AEP that doesn’t really assume anything about the joint distribution of the variables except for log-concavity. There was more in the talk, but I eagerly await the paper.

**Event-triggered Sampling and Reconstruction of Sparse Real-valued Trigonometric Polynomials**

*Neeraj Sharma; Thippur V. Sreenivas*

This was on non-uniform sampling where the sampler tries to detect level crossings of the analog signal and samples at that point — the rate may not be uniform enough to use existing nonuniform sampling techniques. They come up with a method for reconstructing signals which are real-valued trigonometric polynomials with a few nonzero coefficients (e.g. sparse) and it seems to work pretty decently in experiments.

**Removing Sampling Bias in Networked Stochastic Approximation**

*Vivek Borkar; Raaz Dwivedi*

In networked stochastic approximation, the intermittent communication between nodes may mean that the system tracks a different ODE than the one we want. By modifying the method to account for “local clocks” on each edge, we can correct for this, but we end up with new conditions on the step size to make things work. I am pretty excited about this paper, but as usual, my notes were not quite up to getting the juicy bits. That’s what paper reading is for.

**On Asymmetric Insertion and Deletion Errors**

*Ankur A. Kulkarni*

The insertion/deletion channel model is notoriously hard. Ankur proposed a new model where ‘s are “indestructible” — they cannot be inserted or deleted. This asymmetric model leads to new asymptotic bounds on the capacity. I don’t really work on this channel model so I can’t get the finer points of the results, but once nice takeaway was that asymptotically, each indestructible in the codeword lets us correct around a deletion more.

Filed under: Uncategorized ]]>

The lyrics (so far) are:

Entropy is awesome!

Entropy is sum minus p log p

Entropy is awesome!

When you work on I.T.Blockwise error vanishes as n gets bigger

Maximize I X Y

Polarize forever

Let’s party foreverI.I.D.

I get you, you get me

Communicating at capacityEntropy is awesome…

This iteration of the lyrics is due to a number of contributors — truly a group effort. If you want to help flesh out the rest of the song, please feel free to email me and we’ll get a group effort going.

More details on the contest will be forthcoming!

Filed under: Uncategorized ]]>

This talk was on designing Bayesian priors for sparse-PCA problems — the key is to find a prior which induces a low-rank structure on the matrix. The model was something like where is a low-rank matrix and is noise. The previous state of the art is by Babacan et al., a paper which I obviously haven’t read, but the method they propose here (which involved some heavy algebra/matrix factorizations) appears to be competitive in several regimes. Probably more of interest to those working on Bayesian methods…

**Non-Convex Sparse Estimation for Signal Processing**

*David Wipf*

More Bayesian methods! Although David (who I met at ICML) was not trying to say that the priors are particularly “correct,” but rather that the penalty functions that they induce on the problems he is studying actually make sense. More of an algorithmist’s approach, you might say. He set up the problem a bit more generally, to minimize problems of the form

where are some operators. He made the case that convex relaxations of many of these problems, while analytically beautiful, have restrictions which are not satisfied in practice, and indeed they often have poor performance. His approach is via Empirical Bayes, but this leads to non-convex problems. What he can show is that the algorithm he proposes is competitive with any method that tries to separate the error from the “low-rank” constraint, and that the new optimization is “smoother.” I’m sure more details are in his various papers, for those who are interested.

**PCA-HDR: A Robust PCA Based Solution to HDR Imaging**

*Adit Bhardwaj; Shanmuganathan Raman*

My apologies for taking fewer notes on this one, but I don’t know much about HDR imaging, so this was mostly me learning about HDR image processing. There are several different ways of doing HDR, from multiple exposures to flash/no-flash, and so on. The idea is that artifacts introduced by the camera can be modeled using the robust PCA framework and that denoting in HDR imaging may be better using robust PCA. I think that looking at some of the approaches David mentioned may be good in this domain, since it seems unlikely to me that these images will satisfy the conditions necessary for convex relaxations to work…

**On Communication Requirements for Secure Computation**

*Vinod M Prabhakaran*

Vinod showed some information theoretic approaches to understanding how much communication is needed for secure computation protocols like remote oblivious transfer: Xavier has , Yvonne has and Zelda wants , but nobody should be able to infer each other’s values. Feige, Killian, and Naor have a protocol for this, which Vinod and Co. can show is communication-optimal. There were several ingredients here, including cut-set bounds, distribution switching, data processing inequalities, and special bounds for 3-party protocols. More details in his CRYPTO paper (and others).

**Artificial Noise Revisited: When Eve Has More Antennas Than Alice**

*Shuiyin Liu; Yi Hong; Emanuele Viterbo*

In a MIMO wiretap setting, if the receiver has more antennas than the transmitter, then the transmitter can send noise in the nullspace of the channel matrix of the direct channel — as long as the eavesdropper has fewer antennas than the transmitter then secure transmission is possible. In this paper they show that positive secrecy capacity is possible even when the eavesdropper has more antennas, but as the number of eavesdropper antennas grows, the achievable rate goes to . Perhaps a little bit of a surprise here!

Filed under: Uncategorized ]]>

I arrived early enough to catch the tutorials on the first day. There was a 3 hour session in the morning and another on the in afternoon. For the morning I decided to expand my horizons by attending Manoj Gopalkrishnan‘s tutorial on the physics of computation. Manoj focused on the question of how much energy it takes to erase or copy a bit of information. He started with some historical context via von Neumann, Szilard, and Landauer to build a correspondence between familiar information theoretic concepts and their physical counterparts. So in this correspondence, relative entropy is the same as free energy. He then turned to look at what one might call “finite time” thermodynamics. Suppose that you have to apply a control that operates in finite time in order to change a bit. One way to look at this is through controlling the transition probabilities in a two-state Markov chain representing the value of the bit you want to fix. You want to drive the resting state (with stationary distribution to something like within time . At this level I more or less understood what was going on, but since my physics background is pretty poor, I think I missed out on how the physical intuition/constraints impact what control strategies you can choose.

Prasad Santhanam gave the other tutorial, which was a bit more solid ground for me. This was not quite a tutorial on large-alphabet probability estimation, but more directly on universal compression and redundancy calculations. The basic setup is that you have a family of distributions and you don’t know which distribution will generate your data. Based on the data sample you want to do something: estimate some property of the distribution, compress the sample to a size close to its entropy, etc. A class can be weakly or strongly compressible, or *insurable* (which means being able to estimate quantiles), and so on. These problems turn out to be a bit different from each other depending on some topological features of the class. One interesting thing to consider for the machine learners out there this stopping time that you need in some analyses. As you are going along, observing the data and doing your task (estimation, compression, etc) can you tell *from the data* that you are doing well? This has major implications for whether or not an online algorithm can even work the way we want it to, and is something Prasad calls “data-driven compressible.”

I’ll try to write another post or two about the talks I saw as well!

Filed under: Uncategorized ]]>