Bo Brinkman

On the impossibility of dimension reduction in l 1

The Johnson--Lindenstrauss lemma shows that any <i>n</i> points in Euclidean space (i.e., ℝ^<i>n</i> with distances measured under the ℓ<inf>2</inf> norm) may be mapped down to <i>O</i>((log <i>n</i>)/ε²) dimensions such that no pairwise distance is distorted by more than a (1 + ε) factor. Determining whether such dimension reduction is possible in ℓ<inf>1</inf> has been an intriguing open question. We show strong lower bounds for general dimension reduction in ℓ<inf>1</inf>. We give an explicit family of <i>n</i> points in ℓ<inf>1</inf> such that any embedding with constant distortion <i>D</i> requires <i>n</i>^{Ω(1/<i>D</i>²)} dimensions. This proves that there is no analog of the Johnson--Lindenstrauss lemma for ℓ<inf>1</inf>; in fact, embedding with any constant distortion requires <i>n</i>^Ω(1) dimensions. Further, embedding the points into ℓ<inf>1</inf> with (1+ε) distortion requires <i>n</i>^{½−<i>O</i>(ε log(1/ε))} dimensions. Our proof establishes this lower bound for shortest path metrics of series-parallel graphs. We make extensive use of linear programming and duality in devising our bounds. We expect that the tools and techniques we develop will be useful for future investigations of embeddings into ℓ<inf>1</inf>.

Vertex cuts, random walks, and dimension reduction in series-parallel graphs

We consider questions about vertex cuts in graphs, random walks in metric spaces, and dimension reduction in L₁ and L₂; these topics are intimately connected because they can each be reduced to the existence ofvarious families of real-valued Lipschitz maps on certain metric spaces. We view these issues through the lens of shortest-path metricson series-parallel graphs, and we discussthe implications for a variety of well-known open problems. Our main results follow. Every <i>n</i>-point series-parallel metric embeds into l₁^dom with O(√ log n) distortion, matchinga lower bound of Newman and Rabinovich. Our embeddings yield an O(√log n) approximation algorithm for vertex sparsestcut in such graphs, as well as an O(√log k) approximate max-flow/min-vertex-cut theorem for series-parallel instances with<i>k</i> terminals, improving over the O(log n) and O(log k) boundsfor general graphs. Every <i>n</i>-point series-parallel metric embeds withdistortion D into l₁^d with d = n^1/Ω(D²),matching the dimension reduction lower bound of Brinkman andCharikar. There exists a constant C > 0 such that if (X,d) is aseries-parallel metric then for every stationary, reversible Markovchain Z_t_t=0^∞ on X, we have for all t ≥ 0, E[d(Z_t,Z₀)²] ≤ Ct ·, E[d(Z₀,Z₁)²]. More generally, we show thatseries-parallel metrics have Markov type 2. This generalizesa result of Naor, Peres, Schramm, and Sheffield for trees.

An Analysis of Student Privacy Rights in the Use of Plagiarism Detection Systems

Plagiarism detection services are a powerful tool to help encourage academic integrity. Adoption of these services has proven to be controversial due to ethical concerns about students’ rights. Central to these concerns is the fact that most such systems make permanent archives of student work to be re-used in plagiarism detection. This computerization and automation of plagiarism detection is changing the relationships of trust and responsibility between students, educators, educational institutions, and private corporations. Educators must respect student privacy rights when implementing such systems. Student work is personal information, not the property of the educator or institution. The student has the right to be fully informed about how plagiarism detection works, and the fact that their work will be permanently archived as a result. Furthermore, plagiarism detection should not be used if the permanent archiving of a student’s work may expose him or her to future harm.

The heart of a whistle-blower: a corporate decision-making game for computer ethics classes

I describe a simple game for use in starting a class discussion about corporate decision-making and whistle-blowing. The game allows students to experience the power of managers to influence (for good or bad) the decisions of their underlings, and the counter-balancing powers held by workers.

Ethics and Pervasive Augmented Reality: Some Challenges and Approaches

Google has recently announced Project Glass, a plan to embed computers into eyeglasses. Their vision is that technology should “be there when you need it and get out of your way when you don’t.” The main idea behind this technology is called augmented reality. In this chapter I will introduce the main ideas behind augmented reality, and the most interesting examples (so far) of how it is used. I’m most interested, however, in the ethical challenges this technology will present as it becomes pervasive. I describe and explain two examples; augmented reality as an extension of the home (which has privacy implications), and augmented reality advertising (which has implications for property rights and local government). Along the way I will introduce several techniques and concepts that are useful in analyzing the ethics of new technologies, including pervasive information and computing technology (PICT).

AR in the library: A pilot study of multi-target acquisition usability

Libraries use call numbers to organize their books and enable patrons to locate them. To keep the books in order, library workers conduct a time-consuming and tedious task called “shelf-reading.” Workers look at the call numbers on the spines of each book in the library, one at a time, to make sure they are in the correct places. ShelvAR is an augmented reality shelf-reading system for smart phones that reduces time spent, increases accuracy, and produces an inventory of the books on their shelves as a byproduct. Shelf-reading requires rapid acquisition of many targets (books). Unlike many target acquisition tasks considered in the AR literature, the user is not trying to select a single target from among many. Instead, the user is trying to scan all of the targets, and must be able to easily double-check that none were missed. Our goal is to explore the usability of augmented reality applications for this type of “multiple target acquisition” task. We present the results of a pilot study on the effectiveness of ShelvAR. We demonstrate that individuals with no library experience are just as fast and accurate, when using ShelvAR, as experienced library workers at the shelf-reading task.

Willing to be fooled: Security and autoamputation in augmented reality

What does it mean to trust, or not trust, an augmented reality system? Froma computer security point of view, trust in augmented reality represents a real threat to real people. The fact that augmented reality allows the programmer to tinker with the users senses creates many opportunities for malfeasance. It might be natural to think that if we warn users to be careful it will lower their trust in the system, greatly reducing risk.

Color-via-pattern: distinguishing colors of confusion without affecting perceived brightness

In this poster we describe pilot work on a new visualization technique, Color-via-Pattern (CvP), to help individuals with color deficiency distinguish between colors of confusion, as well as correctly determine relative perceived brightness among all colors. Existing assistive technologies tend to distort the hue and brightness of colors. CvP is designed to address this flaw while being just as effective. Human subject testing was performed to evaluate our approach.

All hands on deck for ACM Ethics: updating the Code, revising enforcement, promoting integrity

The Association for Computing Machinerys Committee on Professional Ethics (COPE) has been charged to execute three major projects over the next two years: updating ACMs Code of Ethics and Professional Conduct, revising the enforcement procedures for the Code, and developing new media to promote integrity in the profession. We cannot do this alone, and we are asking SIGCAS members to volunteer and get involved. We will briefly describe the rationale and plan behind these projects and describe opportunities to get involved.

Caption Placement on an Augmented Reality Head Worn Device

Head worn augmented reality devices present new opportunities for creating assistive devices. This work investigates possible locations for captions for one-on-one conversations from a visibility and usability standpoint. Even in a one-on-one conversation there are interesting design challenges. We investigate 3 main factors: 1) Reference frame for captions (viewport or world coordinates); 2) Relation to the viewer (track the viewers view, or track the speaker); and 3) Relation to speakers face (most desirable vertical offset for captions). In this preliminary work, we argue that world coordinates are preferable (for reference frame), and then present our work in progress on the other factors.