ISIT swag has supermodular utility

Proof by figure.

A letter to the GlobalSIP Technical Program Chairs

I am a big supporter of robust peer review. However, I feel very strongly that issuing short review deadlines weakens the review process and has a negative impact on the quality of research. I had a previous experience with a machine learning conference that assigned 9 papers requiring an in-depth review and was given less than 3 weeks to complete these reviews. I immediately wrote back saying that this was infeasible and the deadline was extended by more than a week, as I recall. It was still hard to get the reviews done on time, but I managed it.

People may think I am being petty here, but I think it is important to not get caught in the dilemma of “phone it in and get it done by the deadline” and “pull some all-nighters to get it done right.”

I regret to inform you that I must resign from the Technical Program Committee for GlobalSIP because I will be unable to complete reviews required of me in the time required by the conference.

On July 12 at around 11:45 EST I was assigned 12 papers to review for the conference, for a total of around 60 pages of material (including references). The deadline given was “before July 22, 2015 (AoE)” which I take to mean approximately 8 AM EST July 22 given the location international date line. This is around 9 days to review 12 papers.

At that time I responded indicating that given my other responsibilities, I would be unable to review such a large volume of material at such short notice in the given time frame. I received no response.

On July 15 at 12:33 AM I received a second request to review the same papers with a revised deadline of “before July 25, 2015 (AoE)”. That is, 2 days after the initial assignment, the deadline was extended by 3 days.

Given my other professional and personal commitments, I will not be able to provide the level of scrutiny required to review the papers in under two weeks. As it stands, the modest extension covering 3 additional business days is not enough, especially given the delay in issuing the extension. I realize that conference submissions do not entail the same depth of review as a journal paper, but they still take time, and the review requests came quite unexpectedly.

Finally, I recognize that the delay in assignment was caused by “system glitches” (as stated in your email) and is not the fault of the PC chairs. However, the brunt of the effect is faced by the reviewers. Without any prior communication or information regarding the delay in review assignments, I am not able to juggle/move/delay other obligations at such short notice.

Anand D. Sarwate
Assistant Professor
Department of Electrical and Computer Engineering
Rutgers, The State University of New Jersey
asarwate@ece.rutgers.edu
http://www.ece.rutgers.edu/~asarwate/

Chicago performance group Manual Cinema has a performance running for another week down in the Financial District at 3LD Arts and Technology Center. It’s a co-production with The Tank, a great nonprofit that supports the development of new works. The show uses 4 overhead projectors, a live band, and actors to make a dialogue-free live-created shadow-play animation, complete with sound effects. The overall aesthetics reminded me of Limbo, an independent game, although less zombie-filled. The story is about two sisters, Ada, and Ava. Ava passes away and Ada has to cope with the grief of losing someone so close to her. We go through her memories of their time together and her fantasies light and dark as she mourns and perhaps begins to heal.

I don’t have too much to say except to recommend it highly. If I had one critique it would be that I wanted the story to be more surprising, or revelatory. The medium is at the same time familiar (animation) and new (live performance). They can show so many things and use metaphor (sometimes a bit heavy-handed in a 1940s way) in ways that a conventional play with dialogue would be hard-pressed to do. I wanted to learn something new about grieving, and afterwards I felt like I hadn’t. But then again, I’m still thinking about it, so perhaps I cannot yet put what I have learned into words.

ISIT 2015 : statistics and learning

The advantage of flying to Hong Kong from the US is that the jet lag was such that I was actually more or less awake in the mornings. I didn’t take such great notes during the plenaries, but they were rather enjoyable, and I hope that the video will be uploaded to the ITSOC website soon.

There were several talks on entropy estimation in various settings that I did not take great notes on, to wit:

• OPTIMAL ENTROPY ESTIMATION ON LARGE ALPHABETS VIA BEST POLYNOMIAL APPROXIMATION (Yihong Wu, Pengkun Yang, University Of Illinois, United States)
• DOES DIRICHLET PRIOR SMOOTHING SOLVE THE SHANNON ENTROPY ESTIMATION PROBLEM? (Yanjun Han, Tsinghua University, China; Jiantao Jiao, Tsachy Weissman, Stanford University, United States)
• ADAPTIVE ESTIMATION OF SHANNON ENTROPY (Yanjun Han, Tsinghua University, China; Jiantao Jiao, Tsachy Weissman, Stanford University, United States)

I would highly recommend taking a look for those who are interested in this problem. In particular, it looks like we’re getting towards more efficient entropy estimators in difficult settings (online, large alphabet), which is pretty exciting.

QUICKEST LINEAR SEARCH OVER CORRELATED SEQUENCES
Javad Heydari, Ali Tajer, Rensselaer Polytechnic Institute, United States
This talk was about hypothesis testing where the observer can control the samples being taken by traversing a graph. We have an $n$-node graph (c.f. a graphical model) representing the joint distribution on $n$ variables. The data generated is i.i.d. across time according to either $F_0$ or $F_1$. At each time you get to observe the data from only one node of the graph. You can either observe the same node as before, explore by observing a different node, or make a decision about whether the data from from $F_0$ or $F_1$. By adopting some costs for different actions you can form a dynamic programming solution for the search strategy but it’s pretty heavy computationally. It turns out the optimal rule for switching has a two-threshold structure and can be quite a bit different than independent observations when the correlations are structured appropriately.

MISMATCHED ESTIMATION IN LARGE LINEAR SYSTEMS
Yanting Ma, Dror Baron, North Carolina State University, United States; Ahmad Beirami, Duke University, United States
The mismatch studied in this paper is a mismatch in the prior distribution for a sparse observation problem $y = Ax + \sigma_z z$, where $x \sim P$ (say a Bernoulli-Gaussian prior). The question is what happens when we do estimation assuming a different prior $Q$. The main result of the paper is an analysis of the excess MSE using a decoupling principle. Since I don’t really know anything about the replica method (except the name “replica method”), I had a little bit of a hard time following the talk as a non-expert, but thankfully there were a number of pictures and examples to help me follow along.

SEARCHING FOR MULTIPLE TARGETS WITH MEASUREMENT DEPENDENT NOISE
Yonatan Kaspi, University of California, San Diego, United States; Ofer Shayevitz, Tel-Aviv University, Israel; Tara Javidi, University of California, San Diego, United States
This was another search paper, but this time we have, say, $K$ targets $W_1, W_2, \ldots, W_K$ uniformly distributed in the unit interval, and what we can do is query at each time $n$ a set $S_n \subseteq [0,1]$ and get a response $Y_n = X_n \oplus Z_n$ where $X_n = \mathbf{1}( \exists W_k \in S_n )$ and $Z_n \sim \mathrm{Bern}( \mu(S_n) + b )$ where $\mu$ is the Lebesgue measure. So basically you can query a set and you get a noisy indicator of whether you hit any targets, where the noise depends on the size of the set you query. At some point $\tau$ you stop and guess the target locations. You are $(\epsilon,\delta)$ successful if the probability that you are within $\delta$ of each target is less than $\epsilon$. The targeting rate is the limit of $\log(1/\delta) / \mathbb{E}[\tau]$ as $\epsilon,\delta \to 0$ (I’m being fast and loose here). Clearly there are some connections to group testing and communication with feedback, etc. They show there is a significant gap between the adaptive and nonadaptive rate here, so you can find more targets if you can adapt your queries on the fly. However, since rate is defined for a fixed number of targets, we could ask how the gap varies with $K$. They show it shrinks.

ON MODEL MISSPECIFICATION AND KL SEPARATION FOR GAUSSIAN GRAPHICAL MODELS
Varun Jog, University of California, Berkeley, United States; Po-Ling Loh, University of Pennsylvania, United States
The graphical model for jointly Gaussian variables has no edge between nodes $i$ and $j$ if the corresponding entry $(\Sigma^{-1})_{ij} = 0$ in the inverse covariance matrix. They show a relationship between the KL divergence of two distributions and their corresponding graphs. The divergence is lower bounded by a constant if they differ in a single edge — this indicates that estimating the edge structure is important when estimating the distribution.

CONVERSES FOR DISTRIBUTED ESTIMATION VIA STRONG DATA PROCESSING INEQUALITIES
Aolin Xu, Maxim Raginsky, University of Illinois at Urbana–Champaign, United States
Max gave a nice talk on the problem of minimizing an expected loss $\mathbb{E}[ \ell(W, \hat{W}) ]$ of a $d$-dimensional parameter $W$ which is observed noisily by separate encoders. Think of a CEO-style problem where there is a conditional distribution $P_{X|W}$ such that the observation at each node is a $d \times n$ matrix whose columns are i.i.d. and where the $j$-th row is i.i.d. according to $P_{X|W_j}$. Each sensor gets independent observations from the same model and can compress its observations to $b$ bits and sends it over independent channels to an estimator (so no MAC here). The main result is a lower bound on the expected loss as s function of the number of bits latex $b$, the mutual information between $W$ and the final estimate $\hat{W}$. The key is to use the strong data processing inequality to handle the mutual information — the constants that make up the ratio between the mutual informations is important. I’m sure Max will blog more about the result so I’ll leave a full explanation to him (see what I did there?)

More on Shannon theory etc. later!