arXiv subjects

Siddhartha Sen

Publications and source records attributed to Siddhartha Sen.

At least 37 records · Page 2Linked to original sources

arXiv2022

Learning Models of Individual Behavior in Chess

AI systems that can capture human-like behavior are becoming increasingly useful in situations where humans may want to learn from these systems, collaborate with them, or engage with them as partners for an extended duration. In order to develop human-oriented AI systems, the problem of predicting human actions -- as opposed to predicting optimal actions -- has received considerable attention. Existing work has focused on capturing human behavior in an aggregate sense, which potentially limits the benefit any particular individual could gain from interaction with these systems. We extend this line of work by developing highly accurate predictive models of individual human behavior in chess. Chess is a rich domain for exploring human-AI interaction because it combines a unique set of properties: AI systems achieved superhuman performance many years ago, and yet humans still interact with them closely, both as opponents and as preparation tools, and there is an enormous corpus of recorded data on individual player games. Starting with Maia, an open-source version of AlphaZero trained on a population of human players, we demonstrate that we can significantly improve prediction accuracy of a particular player's moves by applying a series of fine-tuning methods. Furthermore, our personalized models can be used to perform stylometry -- predicting who made a given set of moves -- indicating that they capture human decision-making at an individual level. Our work demonstrates a way to bring AI systems into better alignment with the behavior of individual people, which could lead to large improvements in human-AI interaction.

cs.AI↗

arXiv2021

Sayer: Using Implicit Feedback to Optimize System Policies

We observe that many system policies that make threshold decisions involving a resource (e.g., time, memory, cores) naturally reveal additional, or implicit feedback. For example, if a system waits X min for an event to occur, then it automatically learns what would have happened if it waited <X min, because time has a cumulative property. This feedback tells us about alternative decisions, and can be used to improve the system policy. However, leveraging implicit feedback is difficult because it tends to be one-sided or incomplete, and may depend on the outcome of the event. As a result, existing practices for using feedback, such as simply incorporating it into a data-driven model, suffer from bias. We develop a methodology, called Sayer, that leverages implicit feedback to evaluate and train new system policies. Sayer builds on two ideas from reinforcement learning -- randomized exploration and unbiased counterfactual estimators -- to leverage data collected by an existing policy to estimate the performance of new candidate policies, without actually deploying those policies. Sayer uses implicit exploration and implicit data augmentation to generate implicit feedback in an unbiased form, which is then used by an implicit counterfactual estimator to evaluate and train new policies. The key idea underlying these techniques is to assign implicit probabilities to decisions that are not actually taken but whose feedback can be inferred; these probabilities are carefully calculated to ensure statistical unbiasedness. We apply Sayer to two production scenarios in Azure, and show that it can evaluate arbitrary policies accurately, and train new policies that outperform the production policies.

cs.LG↗

arXiv2021

Hydrogen Atom: Its Spectrum and Degeneracy Importance of the Laplace-Runge-Lenz Vector

Consider the problem: why does the bound state spectrum $E(n) < 0$, of hydrogen atom Hamiltonian $H$ have more degenerate eigenstates than those required by rotational symmetry? The answer is well known and was demonstrated by Pauli. It is due to an additional conserved vector, $\vec A$, of $H$, called the Laplace-Runge-Lenz vector, that was first discovered for planetary orbits. However, surprisingly, a direct link between degenerate eigenstates of $H$ and the physical labels that describe them is missing. To provide such a link requires, as we show, solving a subtle problem of self adjoint operators. In our discussions we address a number of conceptual historical aspects regarding hydrogen atom that also include a careful discussion of both the classical as well as the quantum vector $\vec A$.

physics.atom-ph↗

arXiv2020

SENSEI: Aligning Video Streaming Quality with Dynamic User Sensitivity

This paper aims to improve video streaming by leveraging a simple observation: users are more sensitive to low quality in certain parts of a video than in others. For instance, rebuffering during key moments of a sports video (e.g., before a goal is scored) is more annoying than rebuffering during normal gameplay. Such dynamic quality sensitivity, however, is rarely captured by current approaches, which predict QoE (quality-of-experience) using one-size-fits-all heuristics that are too simplistic to understand the nuances of video content. Instead of proposing yet another heuristic, we take a different approach: we run a separate crowdsourcing experiment for each video to derive users' quality sensitivity at different parts of the video. Of course, the cost of doing this at scale can be prohibitive, but we show that careful experiment design combined with a suite of pruning techniques can make the cost negligible compared to how much content providers invest in content generation and distribution. Our ability to accurately profile time-varying user sensitivity inspires a new approach: dynamically aligning higher (lower) quality with higher (lower) sensitivity periods. We present a new video streaming system called SENSEI that incorporates dynamic quality sensitivity into existing quality adaptation algorithms. We apply SENSEI to two state-of-the-art adaptation algorithms. SENSEI can take seemingly unusual actions: e.g., lowering bitrate (or initiating a rebuffering event) even when bandwidth is sufficient so that it can maintain a higher bitrate without rebuffering when quality sensitivity becomes higher in the near future. Compared to state-of-the-art approaches, SENSEI improves QoE by 15.1% or achieves the same QoE with 26.8% less bandwidth on average.

cs.NI↗

arXiv2020

A New Abstraction for Internet QoE Optimization

A perennial quest in networking research is how to achieve higher quality of experience (QoE) for users without incurring more resources. This work revisits an important yet often overlooked piece of the puzzle: what should the QoE abstraction be? A QoE abstraction is a representation of application quality that describes how decisions affect QoE. The conventional wisdom has relied on developing hand-crafted quality metrics (e.g., video rebuffering events, web page loading time) that are specialized to each application, content, and setting. We argue that in many cases, it maybe fundamentally hard to capture a user's perception of quality using a list of handcrafted metrics, and that expanding the metric list may lead to unnecessary complexity in the QoE model without a commensurate gain. Instead, we advocate for a new approach based on a new QoE abstraction called visual rendering. Rather than a list of metrics, we model the process of quality perception as a user watching a continuous "video" (visual rendering) of all the pixels on their screen. The key advantage of visual rendering is that it captures the full experience of a user with the same abstraction for all applications. This new abstraction opens new opportunities (e.g., the possibility of end-to-end deep learning models that infer QoE directly from a visual rendering) but it also gives rise to new research challenges (e.g., how to emulate the effect on visual rendering of an application decision). This paper makes the case for visual rendering as a unifying abstraction for Internet QoE and outlines a new research agenda to unleash its opportunities.

cs.NI↗

arXiv2020

Aligning Superhuman AI with Human Behavior: Chess as a Model System

As artificial intelligence becomes increasingly intelligent---in some cases, achieving superhuman performance---there is growing potential for humans to learn from and collaborate with algorithms. However, the ways in which AI systems approach problems are often different from the ways people do, and thus may be uninterpretable and hard to learn from. A crucial step in bridging this gap between human and artificial intelligence is modeling the granular actions that constitute human behavior, rather than simply matching aggregate human performance. We pursue this goal in a model system with a long history in artificial intelligence: chess. The aggregate performance of a chess player unfolds as they make decisions over the course of a game. The hundreds of millions of games played online by players at every skill level form a rich source of data in which these decisions, and their exact context, are recorded in minute detail. Applying existing chess engines to this data, including an open-source implementation of AlphaZero, we find that they do not predict human moves well. We develop and introduce Maia, a customized version of Alpha-Zero trained on human chess games, that predicts human moves at a much higher accuracy than existing engines, and can achieve maximum accuracy when predicting decisions made by players at a specific skill level in a tuneable way. For a dual task of predicting whether a human will make a large mistake on the next move, we develop a deep neural network that significantly outperforms competitive baselines. Taken together, our results suggest that there is substantial promise in designing artificial intelligence systems with human collaboration in mind by first accurately modeling granular human decision-making.

cs.AI↗

arXiv2019

MLSys: The New Frontier of Machine Learning Systems

Machine learning (ML) techniques are enjoying rapidly increasing adoption. However, designing and implementing the systems that support ML models in real-world deployments remains a significant obstacle, in large part due to the radically different development and deployment profile of modern ML methods, and the range of practical concerns that come with broader adoption. We propose to foster a new systems machine learning research community at the intersection of the traditional systems and ML communities, focused on topics such as hardware systems for ML, software systems for ML, and ML optimized for metrics beyond predictive accuracy. To do this, we describe a new conference, MLSys, that explicitly targets research at the intersection of systems and machine learning with a program committee split evenly between experts in systems and ML, and an explicit focus on topics at the intersection of the two.

cs.LG↗

arXiv2019

Disaggregation and the Application

This paper examines disaggregated data center architectures from the perspective of the applications that would run on these data centers, and challenges the abstractions that have been proposed to date. In particular, we argue that operating systems for disaggregated data centers should not abstract disaggregated hardware resources, such as memory, compute, and storage away from applications, but should instead give them information about, and control over, these resources. To this end, we propose additional OS abstractions and interfaces for disaggregation and show how they can improve data transfer in data parallel frameworks and speed up failure recovery in replicated, fault-tolerant applications. This paper studies the technical challenges in providing applications with this additional functionality and advances several preliminary proposals to overcome these challenges.

cs.OS↗

arXiv2019

Holography in de Sitter and anti-de Sitter Spaces and Gel'fand Graev Radon transform

Bulk reconstruction formulas similar to HKLL are obtained for de Sitter and anti-de Sitter spaces as the inverse Gel'fand Graev Radon transform. While these generalize our previous result on the Euclidean anti-de Sitter space, their validity in here is restricted only to odd dimensions in both instances. The exact Wightman function for the de Sitter space is then derived analytically. The GGR transform fixes the coefficient of the Wightman function. For the anti-de Sitter space it is shown that a reconstruction formula exists for the case of time-like boundary as well. The restriction on the domain of integration on the boundary is derived. As a special case, we point out that the formula is valid for the BTZ black hole as well.

hep-th↗

arXiv2018

Poor Video Streaming Performance Explained (and Fixed)

HTTP-based video streaming is a key application on the Internet today, comprising the majority of Internet traffic today. Yet customers remain dissatisfied with video quality, resulting in lost revenue for content providers. Recent studies have blamed this on the adaptive bitrate selection (ABR) algorithm used by client players, claiming it interacts poorly with TCP when the video buffer is full, which causes it to underestimate available network bandwidth. We show that the root cause of the problem lies in the data plane, and that even a perfect control plane (ABR) algorithm is not enough to guarantee video flows their fair share of network bandwidth. Namely, it is the sequential download of (small) video segments that is at fault, as they disrupt the normal interaction between TCP congestion control and router queue occupancy. We analytically derive the throughput of a video flow as a function of download size and network conditions, and use this to develop an adaptive algorithm for selecting the download size. Combined with pipelining, our approach achieves near-optimal throughput and fast bitrate adaptation, regardless of the control plane algorithm. We implement our approach as a DASH video player called Sprint, and evaluate it against state-of-the-art proposals from the literature as well as deployed players from Netflix, Youtube, Hulu, and Amazon. Sprint consistently achieves above 90% of its fair-share throughput, while the previous state-of-the-art exhibits high variability (e.g., from 31% to close to fair share depending on the network conditions). Industry players often achieve below 50% of their fair share.

cs.NI↗

arXiv2018

Measuring Consciousness

A measurement based formula for consciousness, C, as a function of time t, is constructed. The formula depends on identifying a natural relevant self-generated, time-dependent dynamical process inherent in any entity. For human beings the relevant dynamical process identified is the ensemble of brain waves, observed in EEG measurements, that are represented in the model by their measured time dependent correlation functions. These correlation functions define the accessible dynamical state of the brain at any moment of time. From them a time dependent probability function, P(t), is extracted by using a mathematical identity. According to information theory, -P(t) log P(t), is a measure of the information contained in the brain waves. Consciousness, C, is defined by this information theory formula: it is not localized, does not depend on specific hardwire details of the brain, but reflects the information content present in brain waves. Justifications, based on observational evidence, are given for the formula and it is shown that C reflects the degree of "awareness" that a person has at a given moment of time. The model explains the observed time delay between when a brain wave is seen to initiate an action and when there is awareness that the action has been initiated, in terms of the way brain waves processes information. Some testable consequences of the model, including the role dreaming sleep plays in long term memory storage, are discussed. It is also shown that non living entities have C=0.

q-bio.NC↗

arXiv2017

Bulk reconstruction in AdS and Gel'fand-Graev-Radon Transform

The bulk reconstruction formula for a Euclidean anti-de Sitter space is directly related to the inverse of the Gel'fand-Graev-Radon transform. Correlation functions of a conformal scalar field theory in the boundary are thereby related to correlation functions of a self-interacting scalar field theory in the bulk at different loop orders.

hep-th↗

arXiv2017

Pyramid: Enhancing Selectivity in Big Data Protection with Count Featurization

Protecting vast quantities of data poses a daunting challenge for the growing number of organizations that collect, stockpile, and monetize it. The ability to distinguish data that is actually needed from data collected "just in case" would help these organizations to limit the latter's exposure to attack. A natural approach might be to monitor data use and retain only the working-set of in-use data in accessible storage; unused data can be evicted to a highly protected store. However, many of today's big data applications rely on machine learning (ML) workloads that are periodically retrained by accessing, and thus exposing to attack, the entire data store. Training set minimization methods, such as count featurization, are often used to limit the data needed to train ML workloads to improve performance or scalability. We present Pyramid, a limited-exposure data management system that builds upon count featurization to enhance data protection. As such, Pyramid uniquely introduces both the idea and proof-of-concept for leveraging training set minimization methods to instill rigor and selectivity into big data management. We integrated Pyramid into Spark Velox, a framework for ML-based targeting and personalization. We evaluate it on three applications and show that Pyramid approaches state-of-the-art models while training on less than 1% of the raw data.

cs.CR↗

arXiv2017

Making Contextual Decisions with Low Technical Debt

Applications and systems are constantly faced with decisions that require picking from a set of actions based on contextual information. Reinforcement-based learning algorithms such as contextual bandits can be very effective in these settings, but applying them in practice is fraught with technical debt, and no general system exists that supports them completely. We address this and create the first general system for contextual learning, called the Decision Service. Existing systems often suffer from technical debt that arises from issues like incorrect data collection and weak debuggability, issues we systematically address through our ML methodology and system abstractions. The Decision Service enables all aspects of contextual bandit learning using four system abstractions which connect together in a loop: explore (the decision space), log, learn, and deploy. Notably, our new explore and log abstractions ensure the system produces correct, unbiased data, which our learner uses for online learning and to enable real-time safeguards, all in a fully reproducible manner. The Decision Service has a simple user interface and works with a variety of applications: we present two live production deployments for content recommendation that achieved click-through improvements of 25-30%, another with 18% revenue lift in the landing page, and ongoing applications in tech support and machine failure handling. The service makes real-time decisions and learns continuously and scalably, while significantly lowering technical debt.

cs.LG↗

arXiv2017

Emergence of Coherent Nanoscale Structures in Carbon Nantotubes

Recently unusual properties of water in single-walled carbon nanotubes (CNT) with diameters ranging from 1.05 nm to 1.52 nm were observed. It was found that water in the CNT remains in an ice-like phase even when the temperature ranges between 105 - 151 C and 87 - 117 C for CNTs with diameters 1.05 nm and 1.06 nm respectively. Apart from the high freezing points, the solid-liquid phase transition temperature was found to be strongly sensitive to the CNT diameter. In this paper we show that water in such CNT's can admit coherent nanoscale structures provided certain conditions are met. The formation of such coherent structures allows for high values of solid-liquid phase transition temperatures that are in qualitative agreement with the empirical data. The model also predicts that the phase transition temperature scales inversely with the square of the effective radius available for the water flow within the CNT. This is consistent with the observed sensitive dependence of the solid-liquid phase transition temperature on the CNT diameter.

cond-mat.mes-hall↗

arXiv2015

An Induced Representation Method for Studying the Stability of Saturn's Ring

MacKey constructed a set of induced representation character formulas to calculate the sum of different powers of the eigenvalues of a matrix with finite group symmetry. We show that these results of MacKey can be used to derive the stability condition, m > Cn3, of Maxwell using his symmetric model for a ring of Saturn, as a collection of n identical unit mass particles revolving round Saturn, which has mass m, in a circular orbit.

math-ph↗

arXiv2015

Coulomb screening in graphene with topological defects

We analyze the screening of an external Coulomb charge in gapless graphene cone, which is taken as a prototype of a topological defect. In the subcritical regime, the induced charge is calculated using both the Green's function and the Friedel sum rule. The dependence of the polarization charge on the Coulomb strength obtained from the Green's function clearly shows the effect of the conical defect and indicates that the critical charge itself depends on the sample topology. Similar analysis using the Friedel sum rule indicates that the two results agree for low values of the Coulomb charge but differ for the higher strengths, especially in the presence of the conical defect. For a given subcritical charge, the transport cross-section has a higher value in the presence of the conical defect. In the supercritical regime we show that the coefficient of the power law tail of polarization charge density can be expressed as a summation of functions which vary log periodically with the distance from the Coulomb impurity. The period of variation depends on the conical defect. In the presence of the conical defect, the Fano resonances begin to appear in the transport cross-section for a lower value of the Coulomb charge. For both sub and supercritical regime we derive the dependence of LDOS on the conical defect. The effects of generalized boundary condition on the physical observables are also discussed.

cond-mat.mes-hall↗

arXiv2014

On the stability and magnetic properties of surface nanobubbles in water

A model for gas nanobubbles is proposed in which their remarkable stability is explained as due to the presence of a qualitatively different form of water covering the nanobubble surface layer which leads to a reduction of the diffusion coefficient by a factor of 10^9. It is shown that this new form of water is created by the interaction between the electrons of water molecules with the zero point vacuum electromagnetic field. The model gives an estimate for the life time of surface nanobubbles, explains why they are not influenced by surfactants and predicts that they should exhibit nonlinear paramagnetism.

cond-mat.mes-hall↗