@phdthesis{26,
abstract = {Expression of genes is a fundamental molecular phenotype that is subject to evolution by different types of mutations. Both the rate and the effect of mutations may depend on the DNA sequence context of a particular gene or a particular promoter sequence. In this thesis I investigate the nature of this dependence using simple genetic systems in Escherichia coli. With these systems I explore the evolution of constitutive gene expression from random starting sequences at different loci on the chromosome and at different locations in sequence space. First, I dissect chromosomal neighborhood effects that underlie locus-dependent differences in the potential of a gene under selection to become more highly expressed. Next, I find that the effects of point mutations in promoter sequences are dependent on sequence context, and that an existing energy matrix model performs poorly in predicting relative expression of unrelated sequences. Finally, I show that a substantial fraction of random sequences contain functional promoters and I present an extended thermodynamic model that predicts promoter strength in full sequence space. Taken together, these results provide new insights and guides on how to integrate information on sequence context to improve our qualitative and quantitative understanding of bacterial gene expression, with implications for rapid evolution of drug resistance, de novo evolution of genes, and horizontal gene transfer.},
author = {Steinrück, Magdalena},
pages = {109},
publisher = {IST Austria},
title = {{The influence of sequence context on the evolution of bacterial gene expression}},
doi = {10.15479/AT:ISTA:th1059},
year = {2018},
}
@phdthesis{324,
abstract = {Neuronal networks in the brain consist of two main types of neuron, glutamatergic principal neurons and GABAergic interneurons. Although these interneurons only represent 10–20% of the whole population, they mediate feedback and feedforward inhibition and are involved in the generation of high-frequency network oscillations. A hallmark functional property of GABAergic interneurons, especially of the parvalbumin‑expressing (PV+) subtypes, is the speed of signaling at their output synapse across species and brain regions. Several molecular and subcellular factors may underlie the submillisecond signaling at GABAergic synapses. Such as the selective use of P/Q type Ca2+ channels and the tight coupling between Ca2+ channels and Ca2+ sensors of exocytosis. However, whether the molecular identity of the release sensor contributes to these signaling properties remains unclear. Besides, these interneurons are mainly show depression in response to train of stimuli. How could they keep sufficient release to control the activity of postsynaptic principal neurons during high network activity, is largely elusive. For my Ph.D. work, we firstly examined the Ca2+ sensor of exocytosis at the GABAergic basket cell (BC) to Purkinje cell (PC) synapse in the cerebellum. Immunolabeling suggested that BC terminals selectively expressed synaptotagmin 2 (Syt2), whereas synaptotagmin 1 (Syt1) was enriched in excitatory terminals. Genetic elimination of Syt2 reduced action potential-evoked release to ~10% compared to the wild-type control, identifying Syt2 as the major Ca2+ sensor at BC‑PC synapses. Differential adenovirus-mediated rescue revealed Syt2 triggered release with shorter latency and higher temporal precision, and mediated faster vesicle pool replenishment than Syt1. Furthermore, deletion of Syt2 severely reduced and delayed disynaptic inhibition following parallel fiber stimulation. Thus, the selective use of Syt2 as the release sensor at BC–PC synapse ensures fast feedforward inhibition in cerebellar microcircuits. Additionally, we tested the function of another synaptotagmin member, Syt7, for inhibitory synaptic transmission at the BC–PC synapse. Syt7 is thought to be a Ca2+ sensor that mediates asynchronous transmitter release and facilitation at synapses. However, it is strongly expressed in fast-spiking, PV+ GABAergic interneurons and the output synapses of these neurons produce only minimal asynchronous release and show depression rather than facilitation. How could Syt7, a facilitation sensor, contribute to the depressed inhibitory synaptic transmission needs to be further investigated and understood. Our results indicated that at the BC–PC synapse, Syt7 contributes to asynchronous release, pool replenishment and facilitation. In combination, these three effects ensure efficient transmitter release during high‑frequency activity and guarantee frequency independence of inhibition. Taken together, our results confirmed that Syt2, which has the fastest kinetic properties among all synaptotagmin members, is mainly used by the inhibitory BC‑PC synapse for synaptic transmission, contributing to the speed and temporal precision of transmitter release. Furthermore, we showed that Syt7, another highly expressed synaptotagmin member in the output synapses of cerebellar BCs, is used for ensuring efficient inhibitor synaptic transmission during high activity.},
author = {Chen, Chong},
pages = {110},
publisher = {IST Austria},
title = {{Synaptotagmins ensure speed and efficiency of inhibitory neurotransmitter release}},
doi = {10.15479/AT:ISTA:th_997},
year = {2018},
}
@phdthesis{83,
abstract = {A proof system is a protocol between a prover and a verifier over a common input in which an honest prover convinces the verifier of the validity of true statements. Motivated by the success of decentralized cryptocurrencies, exemplified by Bitcoin, the focus of this thesis will be on proof systems which found applications in some sustainable alternatives to Bitcoin, such as the Spacemint and Chia cryptocurrencies. In particular, we focus on proofs of space and proofs of sequential work.
Proofs of space (PoSpace) were suggested as more ecological, economical, and egalitarian alternative to the energy-wasteful proof-of-work mining of Bitcoin. However, the state-of-the-art constructions of PoSpace are based on sophisticated graph pebbling lower bounds, and are therefore complex. Moreover, when these PoSpace are used in cryptocurrencies like Spacemint, miners can only start mining after ensuring that a commitment to their space is already added in a special transaction to the blockchain. Proofs of sequential work (PoSW) are proof systems in which a prover, upon receiving a statement x and a time parameter T, computes a proof which convinces the verifier that T time units had passed since x was received. Whereas Spacemint assumes synchrony to retain some interesting Bitcoin dynamics, Chia requires PoSW with unique proofs, i.e., PoSW in which it is hard to come up with more than one accepting proof for any true statement. In this thesis we construct simple and practically-efficient PoSpace and PoSW. When using our PoSpace in cryptocurrencies, miners can start mining on the fly, like in Bitcoin, and unlike current constructions of PoSW, which either achieve efficient verification of sequential work, or faster-than-recomputing verification of correctness of proofs, but not both at the same time, ours achieve the best of these two worlds.},
author = {Abusalah, Hamza M},
pages = {59},
publisher = {IST Austria},
title = {{Proof systems for sustainable decentralized cryptocurrencies}},
doi = {10.15479/AT:ISTA:TH_1046},
year = {2018},
}
@phdthesis{52,
abstract = {In this thesis we will discuss systems of point interacting fermions, their stability and other spectral properties. Whereas for bosons a point interacting system is always unstable this ques- tion is more subtle for a gas of two species of fermions. In particular the answer depends on the mass ratio between these two species. Most of this work will be focused on the N + M model which consists of two species of fermions with N, M particles respectively which interact via point interactions. We will introduce this model using a formal limit and discuss the N + 1 system in more detail. In particular, we will show that for mass ratios above a critical one, which does not depend on the particle number, the N + 1 system is stable. In the context of this model we will prove rigorous versions of Tan relations which relate various quantities of the point-interacting model. By restricting the N + 1 system to a box we define a finite density model with point in- teractions. In the context of this system we will discuss the energy change when introducing a point-interacting impurity into a system of non-interacting fermions. We will see that this change in energy is bounded independently of the particle number and in particular the bound only depends on the density and the scattering length. As another special case of the N + M model we will show stability of the 2 + 2 model for mass ratios in an interval around one. Further we will investigate a different model of point interactions which was discussed before in the literature and which is, contrary to the N + M model, not given by a limiting procedure but is based on a Dirichlet form. We will show that this system behaves trivially in the thermodynamic limit, i.e. the free energy per particle is the same as the one of the non-interacting system.},
author = {Moser, Thomas},
pages = {115},
publisher = {IST Austria},
title = {{Point interactions in systems of fermions}},
doi = {10.15479/AT:ISTA:th_1043},
year = {2018},
}
@phdthesis{197,
abstract = {Modern computer vision systems heavily rely on statistical machine learning models, which typically require large amounts of labeled data to be learned reliably. Moreover, very recently computer vision research widely adopted techniques for representation learning, which further increase the demand for labeled data. However, for many important practical problems there is relatively small amount of labeled data available, so it is problematic to leverage full potential of the representation learning methods. One way to overcome this obstacle is to invest substantial resources into producing large labelled datasets. Unfortunately, this can be prohibitively expensive in practice. In this thesis we focus on the alternative way of tackling the aforementioned issue. We concentrate on methods, which make use of weakly-labeled or even unlabeled data. Specifically, the first half of the thesis is dedicated to the semantic image segmentation task. We develop a technique, which achieves competitive segmentation performance and only requires annotations in a form of global image-level labels instead of dense segmentation masks. Subsequently, we present a new methodology, which further improves segmentation performance by leveraging tiny additional feedback from a human annotator. By using our methods practitioners can greatly reduce the amount of data annotation effort, which is required to learn modern image segmentation models. In the second half of the thesis we focus on methods for learning from unlabeled visual data. We study a family of autoregressive models for modeling structure of natural images and discuss potential applications of these models. Moreover, we conduct in-depth study of one of these applications, where we develop the state-of-the-art model for the probabilistic image colorization task.},
author = {Kolesnikov, Alexander},
pages = {113},
publisher = {IST Austria},
title = {{Weakly-Supervised Segmentation and Unsupervised Modeling of Natural Images}},
doi = {10.15479/AT:ISTA:th_1021},
year = {2018},
}