-
From Explainable to Interactive AI: A Literature Review on Current Trends in Human-AI Interaction
Authors:
Muhammad Raees,
Inge Meijerink,
Ioanna Lykourentzou,
Vassilis-Javed Khan,
Konstantinos Papangelis
Abstract:
AI systems are increasingly being adopted across various domains and application areas. With this surge, there is a growing research focus and societal concern for actively involving humans in developing, operating, and adopting these systems. Despite this concern, most existing literature on AI and Human-Computer Interaction (HCI) primarily focuses on explaining how AI systems operate and, at tim…
▽ More
AI systems are increasingly being adopted across various domains and application areas. With this surge, there is a growing research focus and societal concern for actively involving humans in developing, operating, and adopting these systems. Despite this concern, most existing literature on AI and Human-Computer Interaction (HCI) primarily focuses on explaining how AI systems operate and, at times, allowing users to contest AI decisions. Existing studies often overlook more impactful forms of user interaction with AI systems, such as giving users agency beyond contestability and enabling them to adapt and even co-design the AI's internal mechanics. In this survey, we aim to bridge this gap by reviewing the state-of-the-art in Human-Centered AI literature, the domain where AI and HCI studies converge, extending past Explainable and Contestable AI, delving into the Interactive AI and beyond. Our analysis contributes to shaping the trajectory of future Interactive AI design and advocates for a more user-centric approach that provides users with greater agency, fostering not only their understanding of AI's workings but also their active engagement in its development and evolution.
△ Less
Submitted 23 May, 2024;
originally announced May 2024.
-
Divide and not forget: Ensemble of selectively trained experts in Continual Learning
Authors:
Grzegorz Rypeść,
Sebastian Cygert,
Valeriya Khan,
Tomasz Trzciński,
Bartosz Zieliński,
Bartłomiej Twardowski
Abstract:
Class-incremental learning is becoming more popular as it helps models widen their applicability while not forgetting what they already know. A trend in this area is to use a mixture-of-expert technique, where different models work together to solve the task. However, the experts are usually trained all at once using whole task data, which makes them all prone to forgetting and increasing computat…
▽ More
Class-incremental learning is becoming more popular as it helps models widen their applicability while not forgetting what they already know. A trend in this area is to use a mixture-of-expert technique, where different models work together to solve the task. However, the experts are usually trained all at once using whole task data, which makes them all prone to forgetting and increasing computational burden. To address this limitation, we introduce a novel approach named SEED. SEED selects only one, the most optimal expert for a considered task, and uses data from this task to fine-tune only this expert. For this purpose, each expert represents each class with a Gaussian distribution, and the optimal expert is selected based on the similarity of those distributions. Consequently, SEED increases diversity and heterogeneity within the experts while maintaining the high stability of this ensemble method. The extensive experiments demonstrate that SEED achieves state-of-the-art performance in exemplar-free settings across various scenarios, showing the potential of expert diversification through data in continual learning.
△ Less
Submitted 19 March, 2024; v1 submitted 18 January, 2024;
originally announced January 2024.
-
Looking through the past: better knowledge retention for generative replay in continual learning
Authors:
Valeriya Khan,
Sebastian Cygert,
Kamil Deja,
Tomasz Trzciński,
Bartłomiej Twardowski
Abstract:
In this work, we improve the generative replay in a continual learning setting to perform well on challenging scenarios. Current generative rehearsal methods are usually benchmarked on small and simple datasets as they are not powerful enough to generate more complex data with a greater number of classes. We notice that in VAE-based generative replay, this could be attributed to the fact that the…
▽ More
In this work, we improve the generative replay in a continual learning setting to perform well on challenging scenarios. Current generative rehearsal methods are usually benchmarked on small and simple datasets as they are not powerful enough to generate more complex data with a greater number of classes. We notice that in VAE-based generative replay, this could be attributed to the fact that the generated features are far from the original ones when mapped to the latent space. Therefore, we propose three modifications that allow the model to learn and generate complex data. More specifically, we incorporate the distillation in latent space between the current and previous models to reduce feature drift. Additionally, a latent matching for the reconstruction and original data is proposed to improve generated features alignment. Further, based on the observation that the reconstructions are better for preserving knowledge, we add the cycling of generations through the previously trained model to make them closer to the original data. Our method outperforms other generative replay methods in various scenarios. Code available at https://github.com/valeriya-khan/looking-through-the-past.
△ Less
Submitted 18 September, 2023;
originally announced September 2023.
-
Hemi-Slant Submanifolds of lcK Manifolds as Warped Products
Authors:
Umar Mohd Khan,
Viqar Azam Khan
Abstract:
We study immersions of a hemi-slant submanifold of lcK manifolds as a warped product with the leaves of the holomorphic (respectively slant) distribution warped and establish characterisation theorems and estimations for the squared length of the second fundamental form in both cases.
We study immersions of a hemi-slant submanifold of lcK manifolds as a warped product with the leaves of the holomorphic (respectively slant) distribution warped and establish characterisation theorems and estimations for the squared length of the second fundamental form in both cases.
△ Less
Submitted 11 January, 2023;
originally announced January 2023.
-
Pointwise Bi-Slant Submanifolds in Locally Conformal Kähler Manifolds Immersed as Warped Products
Authors:
Umar Mohd Khan,
Viqar Azam Khan
Abstract:
We study immersions of pointwise bi-slant submanifolds of locally conformal Kähler manifolds as warped products. In particular, we establish characterisation theorem for a pointwise bi-slant submanifold of a locally conformal Kähler manifold to be immersed as a warped product and show that a necessary condition is that the Lee vector field $B$ is orthogonal to the second factor and the warping fun…
▽ More
We study immersions of pointwise bi-slant submanifolds of locally conformal Kähler manifolds as warped products. In particular, we establish characterisation theorem for a pointwise bi-slant submanifold of a locally conformal Kähler manifold to be immersed as a warped product and show that a necessary condition is that the Lee vector field $B$ is orthogonal to the second factor and the warping function $λ$ satisfies $\text{grad}(\lnλ)=\frac{1}{2}B^T$, where $B^T$ denotes the tangential part of the Lee vector field. We also extend Chen's inequality for the squared length of the second fundamental form to our case and study the corresponding equality case.
△ Less
Submitted 11 January, 2023;
originally announced January 2023.
-
Self-Organizing Teams in Online Work Settings
Authors:
Ioanna Lykourentzou,
Federica Lucia Vinella,
Faez Ahmed,
Costas Papastathis,
Konstantinos Papangelis,
Vassilis-Javed Khan,
Judith Masthoff
Abstract:
As the volume and complexity of distributed online work increases, the collaboration among people who have never worked together in the past is becoming increasingly necessary. Recent research has proposed algorithms to maximize the performance of such teams by grouping workers according to a set of predefined decision criteria. This approach micro-manages workers, who have no say in the team form…
▽ More
As the volume and complexity of distributed online work increases, the collaboration among people who have never worked together in the past is becoming increasingly necessary. Recent research has proposed algorithms to maximize the performance of such teams by grouping workers according to a set of predefined decision criteria. This approach micro-manages workers, who have no say in the team formation process. Depriving users of control over who they will work with stifles creativity, causes psychological discomfort and results in less-than-optimal collaboration results. In this work, we propose an alternative model, called Self-Organizing Teams (SOTs), which relies on the crowd of online workers itself to organize into effective teams. Supported but not guided by an algorithm, SOTs are a new human-centered computational structure, which enables participants to control, correct and guide the output of their collaboration as a collective. Experimental results, comparing SOTs to two benchmarks that do not offer user agency over the collaboration, reveal that participants in the SOTs condition produce results of higher quality and report higher teamwork satisfaction. We also find that, similarly to machine learning-based self-organization, human SOTs exhibit emergent collective properties, including the presence of an objective function and the tendency to form more distinct clusters of compatible teammates.
△ Less
Submitted 15 February, 2021;
originally announced February 2021.
-
Libra: Is it Really about Money?
Authors:
Valerie Khan,
Geoffrey Goodell
Abstract:
The announcement by Facebook that Libra will "deliver on the promise of 'the internet of money'" has drawn the attention of the financial world. Regulators, institutions, and users of financial products have all been prompted to react and, so far, no one managed to convince the association behind Libra to apply the brakes or to convince regulators to stop the project altogether. In this article, w…
▽ More
The announcement by Facebook that Libra will "deliver on the promise of 'the internet of money'" has drawn the attention of the financial world. Regulators, institutions, and users of financial products have all been prompted to react and, so far, no one managed to convince the association behind Libra to apply the brakes or to convince regulators to stop the project altogether. In this article, we propose that Libra might be best seen not as a financial newcomer, but as a critical enabler for Facebook to acquire a new source of personal data. By working with financial regulators seeking to address concerns with money laundering and terrorism, Facebook can position itself for privileged access to high-assurance digital identity information. For this reason, Libra merits the attention of not only financial regulators, but also the state actors that are concerned with reputational risks, the rule of law, public safety, and national defence.
△ Less
Submitted 22 August, 2019; v1 submitted 20 August, 2019;
originally announced August 2019.
-
The Compact Linear Collider (CLIC) - 2018 Summary Report
Authors:
The CLIC,
CLICdp collaborations,
:,
T. K. Charles,
P. J. Giansiracusa,
T. G. Lucas,
R. P. Rassool,
M. Volpi,
C. Balazs,
K. Afanaciev,
V. Makarenko,
A. Patapenka,
I. Zhuk,
C. Collette,
M. J. Boland,
A. C. Abusleme Hoffman,
M. A. Diaz,
F. Garay,
Y. Chi,
X. He,
G. Pei,
S. Pei,
G. Shu,
X. Wang,
J. Zhang
, et al. (671 additional authors not shown)
Abstract:
The Compact Linear Collider (CLIC) is a TeV-scale high-luminosity linear $e^+e^-$ collider under development at CERN. Following the CLIC conceptual design published in 2012, this report provides an overview of the CLIC project, its current status, and future developments. It presents the CLIC physics potential and reports on design, technology, and implementation aspects of the accelerator and the…
▽ More
The Compact Linear Collider (CLIC) is a TeV-scale high-luminosity linear $e^+e^-$ collider under development at CERN. Following the CLIC conceptual design published in 2012, this report provides an overview of the CLIC project, its current status, and future developments. It presents the CLIC physics potential and reports on design, technology, and implementation aspects of the accelerator and the detector. CLIC is foreseen to be built and operated in stages, at centre-of-mass energies of 380 GeV, 1.5 TeV and 3 TeV, respectively. CLIC uses a two-beam acceleration scheme, in which 12 GHz accelerating structures are powered via a high-current drive beam. For the first stage, an alternative with X-band klystron powering is also considered. CLIC accelerator optimisation, technical developments and system tests have resulted in an increased energy efficiency (power around 170 MW) for the 380 GeV stage, together with a reduced cost estimate at the level of 6 billion CHF. The detector concept has been refined using improved software tools. Significant progress has been made on detector technology developments for the tracking and calorimetry systems. A wide range of CLIC physics studies has been conducted, both through full detector simulations and parametric studies, together providing a broad overview of the CLIC physics potential. Each of the three energy stages adds cornerstones of the full CLIC physics programme, such as Higgs width and couplings, top-quark properties, Higgs self-coupling, direct searches, and many precision electroweak measurements. The interpretation of the combined results gives crucial and accurate insight into new physics, largely complementary to LHC and HL-LHC. The construction of the first CLIC energy stage could start by 2026. First beams would be available by 2035, marking the beginning of a broad CLIC physics programme spanning 25-30 years.
△ Less
Submitted 6 May, 2019; v1 submitted 14 December, 2018;
originally announced December 2018.
-
An optimal inequality on warped product semi-slant submanifolds of nearly Kaehler manifolds
Authors:
Siraj Uddin,
Falleh R. Al-Solamy,
V. A. Khan,
Cenap Ozel
Abstract:
Non-existence of warped product semi-slant submanifolds of Kaehler manifolds was proved in [17], it is interesting to find their existence. In this paper, we prove the existence of warped product semi-slant submanifolds of nearly Kaehler manifolds by a characterization. To this end we obtain an inequality for the squared norm of second fundamental form in terms of the warping function and the slan…
▽ More
Non-existence of warped product semi-slant submanifolds of Kaehler manifolds was proved in [17], it is interesting to find their existence. In this paper, we prove the existence of warped product semi-slant submanifolds of nearly Kaehler manifolds by a characterization. To this end we obtain an inequality for the squared norm of second fundamental form in terms of the warping function and the slant angle. The equality case is also discussed.
△ Less
Submitted 24 April, 2014; v1 submitted 14 April, 2014;
originally announced April 2014.
-
Enhanced coupling design of a detuned damped structure for clic
Authors:
A. D'Elia,
R. M. Jones,
A. Grudiev,
T. Higo,
V. F. Khan,
W. Wuensch
Abstract:
The key feature of the improved coupling design in the Damped Detuned Structure (DDS) is focused on the four manifolds. Rectangular geometry slots and rectangular manifolds are used. This results in a significantly stronger coupling to the manifolds compared to the previous design. We describe the new design together with its wakefield damping properties.
The key feature of the improved coupling design in the Damped Detuned Structure (DDS) is focused on the four manifolds. Rectangular geometry slots and rectangular manifolds are used. This results in a significantly stronger coupling to the manifolds compared to the previous design. We describe the new design together with its wakefield damping properties.
△ Less
Submitted 16 May, 2012;
originally announced May 2012.
-
Comparative Wakefield Analysis of a First Prototype of a DDS Structure for CLIC Main Linac
Authors:
A. D'Elia,
R. M. Jones,
V. F. Khan,
A. Grudiev,
W. Wuensch
Abstract:
A Damped Detuned Structure (DDS) for CLIC main linac has been proposed as an alternative to the present baseline design which is based on heavy damping. A first prototype, CLIC_DDS_A, for high power tests has been already designed and is under construction. It is also foreseen to design a further prototype, CLIC_DDS_B, to test both the wakefield suppression and high power performances. Wakefield c…
▽ More
A Damped Detuned Structure (DDS) for CLIC main linac has been proposed as an alternative to the present baseline design which is based on heavy damping. A first prototype, CLIC_DDS_A, for high power tests has been already designed and is under construction. It is also foreseen to design a further prototype, CLIC_DDS_B, to test both the wakefield suppression and high power performances. Wakefield calculations for DDS are, in the early design stage, based on single infinitely periodic cells. Though cell-to-cell interaction is taken into account to calculate the wakefields, it is important to study full structure properties using computational tools. In particular this is fundamental for defining the input parameters for the HOM coupler that is crucial for the performances of DDS. In the following a full analysis of wakefields and impedances based on simulations conducted with finite difference based electromagnetic computer code GdfidL will be presented.
△ Less
Submitted 6 October, 2011;
originally announced October 2011.
-
A High Phase Advance Damped and Detuned Structure for the Main Linacs of Clic
Authors:
V. F. Khan,
A. D'Elia,
A. Grudiev,
R. M. Jones,
W. Wuensch
Abstract:
The main accelerating structures for the CLIC are designed to operate at an average accelerating gradient of 100 MV/m. The accelerating frequency has been optimised to 11.994 GHz with a phase advance of 2π/3 of the main accelerating mode. The moderately damped and detuned structure (DDS) design is being studied as an alternative to the strongly damped WDS design. Both these designs are based on th…
▽ More
The main accelerating structures for the CLIC are designed to operate at an average accelerating gradient of 100 MV/m. The accelerating frequency has been optimised to 11.994 GHz with a phase advance of 2π/3 of the main accelerating mode. The moderately damped and detuned structure (DDS) design is being studied as an alternative to the strongly damped WDS design. Both these designs are based on the nominal accelerating phase advance. Here we explore high phase advance (HPA) structures in which the group velocity of the rf fields is reduced compared to that of standard (2π/3) structures. The electrical breakdown strongly depends on the fundamental mode group velocity. Hence it is expected that electrical breakdown is less likely to occur in the HPA structures. We report on a study of both the fundamental and dipole modes in a CLIC_DDS_HPA structure, designed to operate at 5π/6 phase advance per cell. Higher order dipole modes in both the standard and HPA structures are also studied.
△ Less
Submitted 3 October, 2010;
originally announced October 2010.
-
Recent Progress on a Manifold Damped and Detuned Structure for CLIC
Authors:
V. F. Khan,
A. D'Elia,
A. Grudiev,
R. M. Jones,
W. Wuensch,
R. Zennaro
Abstract:
A damped detuned structure for the main X-band linacs of CLIC is being investigated as an alternative design to the present baseline heavily damped structure. In our earlier designs we studied detuned structures, operating at 11.994 GHz, with a range of dipole bandwidths in order to ensure the structure satisfies beam dynamics and rf breakdown constraints. Here we report on the development of a da…
▽ More
A damped detuned structure for the main X-band linacs of CLIC is being investigated as an alternative design to the present baseline heavily damped structure. In our earlier designs we studied detuned structures, operating at 11.994 GHz, with a range of dipole bandwidths in order to ensure the structure satisfies beam dynamics and rf breakdown constraints. Here we report on the development of a damped and detuned structure which satisfies both constraints. Preparations for high power testing of the structure are also discussed
△ Less
Submitted 1 June, 2010;
originally announced June 2010.
-
Wakefield damping for the CLIC crab cavity
Authors:
P. K. Ambattu,
G. Burt,
A. C. Dexter,
R. G. Carter,
V. Khan,
R. M. Jones,
V. Dolgashev
Abstract:
A crab cavity is required in the CLIC to allow effective head-on collision of bunches at the IP. A high operating frequency is preferred as the deflection voltage required for a given rotation angle and the RF phase tolerance for a crab cavity are inversely proportional to the operating frequency. The short bunch spacing of the CLIC scheme and the high sensitivity of the crab cavity to dipole ki…
▽ More
A crab cavity is required in the CLIC to allow effective head-on collision of bunches at the IP. A high operating frequency is preferred as the deflection voltage required for a given rotation angle and the RF phase tolerance for a crab cavity are inversely proportional to the operating frequency. The short bunch spacing of the CLIC scheme and the high sensitivity of the crab cavity to dipole kicks demand very high damping of the inter-bunch wakes, the major contributor to the luminosity loss of colliding bunches. This paper investigates the nature of the wakefields in the CLIC crab cavity and the possibility of using various damping schemes to suppress them effectively.
△ Less
Submitted 19 June, 2009;
originally announced June 2009.
-
Investigation of an alternate means of wakefield suppression in the main linacs of CLIC
Authors:
V. F. Khan,
R. M. Jones
Abstract:
We report on suppression of long-range wakefields in CLIC accelerating structures. Strong detuning and moderate damping is employed. In these initial design studies we focus on the CLIC_G structure and enforce a moderate Q of 300 and 500. We maintain a dipole bandwidth of approximately 1 GHz as specified from breakdown constraints in a modified structure, CLIC_DDS. The circuit model, which facil…
▽ More
We report on suppression of long-range wakefields in CLIC accelerating structures. Strong detuning and moderate damping is employed. In these initial design studies we focus on the CLIC_G structure and enforce a moderate Q of 300 and 500. We maintain a dipole bandwidth of approximately 1 GHz as specified from breakdown constraints in a modified structure, CLIC_DDS. The circuit model, which facilitates a rapid design of manifolds slot-coupled to the main accelerating cells, is described.
△ Less
Submitted 19 June, 2009;
originally announced June 2009.
-
An alternate design for CLIC main linac wakefield suppression
Authors:
V. F. Khan,
R. M. Jones
Abstract:
The present design of the main accelerating structure for CLIC is based on heavy damping (WDS) with a Q of ~10. The wakefield suppression in this case entails locating the damping materials in relatively close proximity to the accelerating cells. Herein we present an alternate design for the main accelerating structures. We detune the lowest dipole band by prescribing a Gaussian distribution to…
▽ More
The present design of the main accelerating structure for CLIC is based on heavy damping (WDS) with a Q of ~10. The wakefield suppression in this case entails locating the damping materials in relatively close proximity to the accelerating cells. Herein we present an alternate design for the main accelerating structures. We detune the lowest dipole band by prescribing a Gaussian distribution to the cell parameters and consider moderate damping Q~500 to prevent the recoherence of the modes; in this case the damping materials can be located at an extended distance from the accelerating structure. The procedure to achieve a well-damped wakefield is described. Results are presented elucidating the various designs including the current one which is being developed to incorporate r.f. breakdown, pulse surface heating and beam dynamics constraints.
△ Less
Submitted 11 March, 2009;
originally announced March 2009.
-
Beam dynamics and wake-field simulations for the CLIC main linacs
Authors:
V. Khan,
R. M. Jones
Abstract:
The CLIC linear collider aims at accelerating multiple bunches of electrons and positrons and colliding them at a centre of mass energy of 3 TeV. These bunches will be accelerated through X-band linacs, operating at an accelerating frequency of 12 GHz. Each beam readily excites wake-fields within the accelerating cavities of each linac. The transverse components of the wake-fields, if left unche…
▽ More
The CLIC linear collider aims at accelerating multiple bunches of electrons and positrons and colliding them at a centre of mass energy of 3 TeV. These bunches will be accelerated through X-band linacs, operating at an accelerating frequency of 12 GHz. Each beam readily excites wake-fields within the accelerating cavities of each linac. The transverse components of the wake-fields, if left unchecked, can dilute the beam emittance. The present CLIC design relies on heavy damping of these wake-fields in order to ameliorate the effects of the wake-fields on the beam emittance. Here we present initial results on simulations of the long-range wake-fields in these structures and on beam dynamics simulations. In particular, detailed simulations are performed, on emittance dilution due to beams initially injected with realistic offsets from the electrical centre of the cavities.
△ Less
Submitted 15 October, 2008;
originally announced October 2008.
