(The talk will be given in English)

Speaker:     Dr. Doron Blatt
                     DRW Trading, Chicago

Sunday, April 14^th, 2019
15:00 - 16:00

Room 011, Kitot Bldg., Faculty of Engineering

Challenges in Algorithmic Trading: Machine Learning, Optimal Control and Convex Optimization

Abstract

The financial market has evolved in a few short decades from in-person trading in the pits to an estimated 80% of all trading conducted by automated trading systems. This talk will present the key components of the electronic marketplace. The basic building block of electronic trading is the Central Limit Order Book (CLOB), a continuous double auction where supply meets demand and price discovery takes place. I will discuss a model for the CLOB and an optimal control problem that can be formulated and solved to optimize execution strategies. Predicting future prices is the “holy grail” of algorithmic trading. I will give an example where machine learning can be used to make such predictions and discuss the challenges of non-stationary data and “big p little n”. Given predictions, constructing optimal portfolios is an area with many unsolved challenges and research opportunities. The talk will cover other important topics such as preventing over-fitting, online convex optimization and multi-arm bandit problems.

Short Bio

Doron Blatt received his B.Sc. and M.Sc. from Tel Aviv University and Ph.D. from the University of Michigan in Electrical Engineering. He conducted research on signal processing, machine learning, and optimization. Doron joined DRW Trading in 2006 and became a partner in 2011. Today DRW is one of the largest principal trading firms in the world. Over the past 13 years Doron has built and managed DRW’s algorithmic trading group. Doron is returning to Israel to open DRW’s R&D office in Tel Aviv.  

The plasticity of the connectome

Prof. Yaniv Assaf

¹ School of Neurobiology, biochemistry and biophysics, George S. Wise faculty of life sciences

² Sagol School of Neuroscience

Tel Aviv University, Tel Aviv, Israel

At every aspect of our lives – function determines structure. Just as new roads are built between developing cities, network wires are laid to adjust to faster communication demand and social networks are formed under a common goal of individuals, also the brain needs to remodel it's connectome to adapt to the daily and continuous change in functional demands.

The connectome refers to several functional and structural characteristics of brain connectivity that span from the micron level (neural circuits) to the macroscopic level (long scale pathways). This complex network (which includes the white matter but not only) is responsible for the information passage through different regions.

Without the ability to explore the connectome in-vivo, it was traditionally considered to be stable and fixed. Indeed, most effort in white matter research was invested in describing the geographical appearance of the network and the areas it connects. Yet, aside from being a wiring program, does the connectome have an active role in brain plasticity?

Diffusion MRI combined with tractography and graph analysis, allows the exploration of brain connectomics dynamics, in-vivo and on large cohorts. Assuming that the ability of the brain to absorb and store new information relays on the way it is wired, neuroplasticity must depend on connectome characteristics governed by the topology of the network. The relation between network topology and plasticity will be described in presentation

ההרצאה תתקיים ביום ראשון 17.03.2019, בשעה 14:00

בחדר 315, הבניין הרב תחומי, אוניברסיטת תל אביב

​​

(The talk will be given in English)

Speaker:     Prof. Yaakov Bar-Shalom
                     University of Connecticut

Wednesday, March 20^th, 2019
15:00 - 16:00

Room 011, Kitot Bldg., Faculty of Engineering

A Survey of Some Recent Results on the CRLB for Parameter Estimation: Relaxation of Regularity Conditions and the CRLLB for Parameter-Dependent LF

Abstract

This presentation surveys some recent results regarding the Cram´er-Rao Lower Bound (CRLB) — the requirements for its existence and its extension to the situation where the parameter’s likelihood function (LF) support depends on the parameter to be estimated. In the latter case the conventional CRLB does not hold in general. First we revisit the derivation of the CRLB to elucidate the necessary and sufficient conditions needed for the estimation bound. Following this, the situation of parameter- dependent LF support is considered and the Cram´er-Rao-Leibniz Lower Bound (CRLLB) is shown to have an additional term that follows from the Leibniz integral rule, hence the name of this recently obtained bound. This result is the key to provide the estimation bound for the case of uniformly distributed measurement noise over a finite interval. It is also shown that in this case the CRLLB is unattainable. Other examples are also discussed, like the truncated Gaussian measurement noise, in which case the CRLLB is shown to be attained. We also discuss two cases of apparent “super-efficient estimators” where the estimators’ variances are smaller than the conventional CRLB and show that these examples have correct CRLLBs, thus setting straight the myth of “super-efficiency”.