Category Archives: Publications

research collaboration

Call for Collaboration on the paper: “Economics of Human-AI Ecosystem: Value Bias and Lost Utility in Multi-Dimensional Gaps”

The paper on “Economics of Human-AI Ecosystem: Value Bias and Lost Utility in Multi-Dimensional Gaps”[1] was submitted to the AAAI/ACM Conference on AI, Ethics, and Society.

One of the main issues discussed in the paper is the investment in research and knowledge which is a process of putting an effort and paying costs in a few dimensions like time, money, emotional cost and more, with a hope to discover something we did not know. The process of investment in knowledge also involves a risk that is the risk of failure since failure makes all the costs paid for nothing. Real knowledge, most of the time cannot be obtained without failures along the way, since acquiring knowledge and making progress toward some unknown goal is a process of trying, searching, asking and answering and implementing the obtained experience from many failures in a right or better way. In many cases, publishing failures are much more important than publishing success. The logic is simple; if we are doing research aiming to discover an unknown or unfamiliar observations/knowledge, the things that we find out along the way that are wrong are the only things that we actually know (recall, we are on the search for what we do not know). We make a process of learning what is wrong on our way to discover what we do not know -—> So publishing and admitting our failures is the way to help ourselves and the community to improve the way we act. Publishing failures help to choose the right direction and not waste our or our colleagues time on paths with dead-ends.

I know 4 languages, which allows me to communicate with about 50% of the people in the world. However, I have grammar issues and an accent in all of them 🙂

I must admit my problem, I concentrate on the essence, on the core of ideas, which makes the grammar for me an obstacle rather than the cause. This is a paradox since the focus on what is essential practically makes me fail to reach it and deliver my ideas. In my paper on “value-driven landmarks for oversubscription planning,”[2] I treat this paradox. I propose planning and problem-solving (domain independent) approach with a focus on values and landmarks along the way rather than goals. In all the tested benchmarks, in practice, and in theory, relaxing the focus on goals allows us to reach our goals more effectively and more efficiently. This is the paradox of goals.

I have grammar mistakes and language issues that takes the attention of my readers from the essence, the idea, the goal behind my paper and at the end of my words. I fall again and again into the paradox I am trying to solve — I am trying to explain the importance of the process to reach our goals with grammar mistakes, which hold the process and make my goals hard to get.

The feedback regarding my lousy grammar is part of many of my works, and I am familiar with that weakness of mine since school. My teacher in school told me “Your ideas are great, but you have grammar issues,” Thank you, my teacher, now what? What is more important, the idea or grammar? Can you be more specific, please? would you show me the way to fix a mistake and deliver the idea? And if this grammar issue is a bug which is hard to fix what do we do with the ideas?

I am looking for a collaboration to improve and publish that paper, to improve myself and fit into the hard borders of academic research, definitions and grammar strictness. I must say that taking research society perspective, I must stress that academic research community is also in a paradox — the paradox of focus on definitions on the way to discover what not have been defined yet.

Bellow is my paper abstract, version 1 of the paper with reviews, and version 2 of my article that treat some issues mentioned by the reviewers. The reviews raise a problem, but the feedback is not effective with concern to future progress, which makes it hard for me to create a utility from failure.

I fixed some language issues and call for collaboration to make progress with the ideas. I would be happy to have any input, thoughts or guidance to improve the work and target it to the right audience, and would be glad to publish with a mentor\s or a partner\s.

Paper Abstract

In recent years, artificial intelligence (AI) decision-making and autonomous systems became an integrated part of the economy, industry, and society. The evolving economy of the human-AI ecosystem raising concerns regarding the risks and values inherited in AI systems. This paper investigates the dynamics of creation and exchange of values and points out gaps in perception of cost-value, knowledge, space and time dimensions. It shows aspects of value bias in human perception of achievements and costs that encoded in AI systems. It also proposes rethinking hard goals definitions and cost-optimal problem-solving principles in the lens of effectiveness and efficiency in the development of trusted machines. The paper suggests a value-driven with cost awareness strategy and principles for problem-solving and planning of effective research progress to address real-world problems that involve diverse forms of achievements, investments, and survival scenarios.

AIES 2019 REVIEWS

The reviews below are for the first version of the paper that is available here: Version # 1.

Many of the language-related errors fixed in version # 2

*A personal request from reviewer 1, I would love to work with you on that paper, please be my guide and co-author.

—————— REVIEW 1 ———————
PAPER: 234
TITLE: Economics of Human-AI Ecosystem: Value Bias and Lost Utility in Multi-Dimensional Gaps
AUTHORS: Daniel Muller

Overall evaluation: -2 (reject)

———– Overall evaluation ———–
This paper presents a high level analysis the dynamic of exchange of values mostly in the context of autonomous decision making.
One of the most interesting points made in the paper is the introduction of the value-cost tradeoff. The work presented appears to be at a very preliminary a mostly conceptual stage. Moreover, the paper needs major revision from a language stand point. Several sentences are missing verbs and punctuation is not consistent. This make it really hard to following the reasoning behind what is being presented.

———————– REVIEW 2 ———————
PAPER: 234
TITLE: Economics of Human-AI Ecosystem: Value Bias and Lost Utility in Multi-Dimensional Gaps
AUTHORS: Daniel Muller

Overall evaluation: -2 (reject)

———– Overall evaluation ———–
This paper presents itself as very messy, and largely unstructured. Here and there, there are certainly good ideas, but overall the paper is not ready to be accepted. The narrative is not clear, there is no clear context.

———————– REVIEW 3 ———————
PAPER: 234
TITLE: Economics of Human-AI Ecosystem: Value Bias and Lost Utility in Multi-Dimensional Gaps
AUTHORS: Daniel Muller

Overall evaluation: -3 (strong reject)

———– Overall evaluation ———–

This paper is not appropriate for publication. It’s a collection of unrelated, unstructured paragraphs, and it is really hard to understand what it is really about.

[1] D. Muller, “Economics of human-ai ecosystem: value bias and lost utility in multi-dimensional gap,” Arxiv preprint arxiv:1811.06606, 2018.
[Bibtex]
@article{mullerEconomicsHAI2018,
title={Economics of Human-AI Ecosystem: Value Bias and Lost Utility in Multi-Dimensional Gap},
author={Muller, Daniel},
journal={arXiv preprint arXiv:1811.06606},
year={2018}
}
[2] D. Muller and E. Karpas, “Value driven landmarks for oversubscription planning.,” in Icaps, 2018.
[Bibtex]
@inproceedings{muller:Karpas:icaps18,
title={Value Driven Landmarks for Oversubscription Planning.},
author={Muller, Daniel and Karpas, Erez},
booktitle={ICAPS},
year={2018}
}
Frontier Search and Plan Reconstruction in Oversubscription Planning

Check out work on “Frontier Search and Plan Reconstruction in Oversubscription Planning” – in AAAI 2019

Our paper on “Frontier Search and Plan Reconstruction in Oversubscription Planning” [1] will be presented in The Thirty-Third AAAI Conference on Artificial Intelligence (AAAI-19)

Oversubscription planning (OSP) [2] is a problem of choosing an action sequence which reaches a state with a high utility, given a budget for total action cost. This formulation allows to handle situations with under-constrained resources, which do not allow to achieve all possible goal propositions. In optimal OSP, the task is further constrained to finding a path which achieves a state with maximal utility. Best-First-Branch-and-Bound (BFBB) is a heuristic search algorithm which is widely used for solving OSP problems. <em>BFBB</em> relies on an admissible utility-upper-bounding heuristic function (with budget restrictions) \(h : S \times {\mathbb R}^{0+} \rightarrow R\) to estimate the true utility \(h*(s,b)\).
An incremental BFBB search algorithm with landmark-based approximations (<em>inc-compile-and-BFBB</em>) was proposed for OSP heuristic search [3] to address tasks with non-negative and 0-binary utility functions. <em>inc-compile-and-BFBB</em> maintains the best solution so far and a set of reference states, extended with all the non-redundant value-carrying states discovered during the search. Each iteration requires search re-start in order to exploit the new information obtained along the search. Recent work presented a relative estimation of achievements with value-driven landmarks [4] addressing arbitrary additive utility functions, which incrementally improves the best solution so far eliminating the need to maintain a set of reference states.
This paper [1] proposes a <em>progressive frontier search</em> algorithm, which alleviates the computational cost of search restart once new information is acquired. Our technique allows the new search iteration to continue from any state on the frontier of the previous search iteration, leading to improved efficiency of the search. An extended version of this abstract is available online [5].

[1] D. Muller and E. Karpas, “Frontier search and plan reconstruction in oversubscription planning,” in Aaai, 2018.
[Bibtex]
@inproceedings{muller2018frontier,
title={Frontier Search and Plan Reconstruction in Oversubscription Planning},
author={Muller, Daniel and Karpas, Erez},
booktitle={AAAI},
year={2018}
}
[2] D. E. Smith, “Choosing objectives in over-subscription planning.,” in Icaps, 2004, p. 393.
[Bibtex]
@inproceedings{smith:icaps04,
title={Choosing Objectives in Over-Subscription Planning.},
author={Smith, David E},
booktitle={ICAPS},
volume={4},
pages={393},
year={2004}
}
[3] C. Domshlak and V. Mirkis, “Deterministic oversubscription planning as heuristic search: abstractions and reformulations,” Journal of artificial intelligence research, vol. 52, p. 97–169, 2015.
[Bibtex]
@article{mirkis:domshlak:jair15,
title={Deterministic oversubscription planning as heuristic search: Abstractions and reformulations},
author={Domshlak, Carmel and Mirkis, Vitaly},
journal={Journal of Artificial Intelligence Research},
volume={52},
pages={97--169},
year={2015}
}
[4] D. Muller and E. Karpas, “Value driven landmarks for oversubscription planning.,” in Icaps, 2018.
[Bibtex]
@inproceedings{muller:Karpas:icaps18,
title={Value Driven Landmarks for Oversubscription Planning.},
author={Muller, Daniel and Karpas, Erez},
booktitle={ICAPS},
year={2018}
}
[5] D. Muller and E. Karpas, “Value driven landmarks for oversubscription planning,” Technion, Faculty of Industrial Engineering and Management, IE/IS-2018-04, 2018.
[Bibtex]
@techreport{muller2018TRvalue,
title = {Value Driven Landmarks for Oversubscription Planning},
author = {Muller, Daniel and Karpas, Erez},
year = {2018},
institution = {Technion, Faculty of Industrial Engineering and Management},
number = {IE/IS-2018-04}
}
Logistic Task

Check out my new paper on Automated Tactical Decision Planning Model with Strategic Values Guidance for Local Action-Value-Ambiguity”

In many real-world planning problems, action’s impact differs with a place, time and the context in which the action is applied. The same action with the same effects in a different context or states can cause a different change. In actions with incomplete precondition list, that applicable in several states and circumstances, ambiguity regarding the impact of the action is challenging even in small domains. To estimate the real impact of actions, an evaluation of the effect list will not be enough; a relative estimation is more informative and suitable for estimation of action’s real impact. Recent work on Over-subscription Planning (OSP) defined the net utility of action as the net change in the state’s value caused by the action. The notion of net utility of action allows for a broader perspective on value action impact and use for a more accurate evaluation of achievements of the action, considering inter-state and intra-state dependencies. To achieve value-rational decisions in complex reality often requires strategic, high level, planning with a global perspective and values, while many local tactical decisions require real-time information to estimate the impact of actions. This paper proposes an offline action-value structure analysis to exploit the compactly represented informativeness of net utility of actions to extend the scope of planning to value uncertainty scenarios and to provide a real-time value-rational decision planning tool. The result of the offline pre-processing phase is a compact decision planning model representation for flexible, local reasoning of net utility of actions with (offline) value ambiguity. The obtained flexibility is beneficial for the online planning phase and real-time execution of actions with value ambiguity. Our empirical evaluation shows the effectiveness of this approach in domains with value ambiguity in their action-value-structure.

Automated Tactical Decision Planning Model with Strategic Values Guidance for Local Action-Value-Ambiguity

 Automated Tactical Decision Planning Model with Strategic Values Guidance for Local Action-Value-Ambiguity

 

Optimal utility value ambiguity

 Optimal utility value ambiguity

  • D. Muller and E. Karpas, “Automated tactical decision planning model with strategic values guidance for local action-value-ambiguity,” Arxiv preprint arxiv:1811.12917, 2018.
    [Bibtex]
    @article{muller2018tactic,
    title={Automated Tactical Decision Planning Model with Strategic Values Guidance for Local Action-Value-Ambiguity},
    author={Muller, Daniel and Karpas, Erez},
    journal={arXiv preprint arXiv:1811.12917},
    year={2018}
    }
Economics of Human-AI Ecosystem: Value Bias and Lost Utility in Multi-Dimensional Gaps

Check out my new paper on “Economics of Human-AI Ecosystem: Value Bias and Lost Utility in Multi-Dimensional Gaps”

In recent years, artificial intelligence (AI) decision-making and autonomous systems became an integrated part of the economy, industry, and society. The evolving economy of the human-AI ecosystem raising concerns regarding the risks and values inherited in AI systems. This paper [1] investigates the dynamics of creation and exchange of values and points out gaps in perception of cost-value, knowledge, space and time dimensions. It shows aspects of value bias in human perception of achievements and costs that encoded in AI systems. It also proposes rethinking hard goals definitions and cost-optimal problem-solving principles in the lens of effectiveness and efficiency in the development of trusted machines. The paper suggests a value-driven with cost awareness strategy and principles for problem-solving and planning of effective research progress to address real-world problems that involve diverse forms of achievements, investments, and survival scenarios.

 

DIKW hierarchy

The relation of information and knowledge modeled in
the in DIKW hierarchy (data – information – knowledge –
wisdom)

Different dimensions of costs and values

Different dimensions of costs and values

 

Types of utilities and examples for creating positive and negative utility

Types of utilities and examples for creating positive and negative utility

 

[1] D. Muller, “Economics of human-ai ecosystem: value bias and lost utility in multi-dimensional gap,” Arxiv preprint arxiv:1811.06606, 2018.
[Bibtex]
@article{mullerEconomicsHAI2018,
title={Economics of Human-AI Ecosystem: Value Bias and Lost Utility in Multi-Dimensional Gap},
author={Muller, Daniel},
journal={arXiv preprint arXiv:1811.06606},
year={2018}
}

Seminar Slides (2016) on Deterministic Oversubscription Action Planning with General Utility Functions

I started my work on the complex but fascinating Oversubscription planning (OSP) problem at Merch 2014.  In 2016 I have finalized my observations on oversubscription action planning with negative value effects which are captured in the more general term of planning with negative utility functions in my master thesis. Since most of the current work on OSP was concerned with 0-binary and non-negative utility functions, the novel results on planning with negative utility functions were uncomparable with the existing state-of-the-art at the time. Although real-world problems involve negative effects and actions with implicit and explicit negative utilities as in OSP problem it was quite difficult to show along the way the importance of this problem. The result of this work was published only in 2018 in a paper on Value Driven Landmarks for Oversubscription Planning. In my recent paper on “Economics of Human-AI Ecosystem: Value Bias and Lost Utility in Multi-Dimensional Gaps” I discuss the importance of real-world problem solving more deeply.

Two years after this seminar and about three years after writing that thesis, I am still struggling to publish the same old work…

My Seminar abstract from 2016 is available here

Seminar slides from 2016 are available here

Continue Reading Seminar Slides (2016) on Deterministic Oversubscription Action Planning with General Utility Functions

What is it Value Landmarks? Why shall we use them?

Actually, I didn’t find a reason yet why shouldn’t we use them. Computationally cheap, can improve but do not do damage; In the worst case scenario using them will achieve the same results as not using. It just an extra, very relevant information that you can use or not. Make your choice… 🙂
But if you find a reason why not to use them, please let me know, contact me or leave a comment. Thanks!

Here are two examples where value-driven landmarks can lead us to better decisions in a complex (everyday life), oversubscription planning problems with sequential actions.

Bill achieved his goals! Well, it is a matter of perspective… Value Driven Landmarks for Oversubscription Planning show us a wider picture. Process-Oriented flexible planning for online, landmark-based sequential acting planning and decision making with negative utility interactions between variables (which widely appear also in the non-negative scenarios). Synergistic Criteria for a process of improvement defines a window of opportunity to terminate the process with a profit. Sometimes we better do nothing.

The context of negative effects and interactions with targets can be captured with the net utility of actions definition within Value Driven Landmarks for Oversubscription Planning. A sequence of actions that improve utility terminates with a positive net utility value action. The net and gross utility value of action definitions take into account a wider context of achievements, by capturing inter-state dependencies and intra-state dependencies.

For more details on how we find these Value Landmarks check out the introduction to oversubscription planning and our publications.

Oversubscription Planning (OSP) value landmarks

Technical Report for the Paper on Value Driven Landmarks for Oversubscription Planning

A detailed technical report of out ICAPS 2018 paper Value Driven Landmark for Oversubscription Planning is available. In the technical report, we provide detail examples of the theory in the paper. We look closely at the terms of optimality and achievements concerning the complexity of the real-world scenarios. ICAPS 2018 Slides of our presentation along with supplementary material can be found at the publication page.
Starting with the most fundamental question of what additive utility function in OSP problem is, we point out the challenges in multi-valued planning tasks with additive utility setting. We discuss the relationships between state variables and different value assignments to a variable in successive states along a plan. We closely consider negative interactions between state variables with multi-valued (non-zero binary) utility setting, and we show how these negative interactions could occur in tasks with non-negative utility setting.
We treat the OSP task as a process of improvement of the initial state rather than a process of collecting valuable facts is the most basic fundamental of our approach. In contrast to classical planning and partial satisfaction problems where there is one explicit assignment for each variable that is defined as valuable, OSP with additive utility functions allows for each variable to be associated with a set of different utilities. Thus, in the additive utility case, a variable assignment is valuable if its utility is better than the utility at the initial state, where an optimal solution will be the maxima(red circle) l utility over all variables that are {\em mutually consistent}. Therefore, it is easy to see that the concept of {\em improving states} rather than collecting valuable facts is much more suitable for the general case.
In order to capture the properties of the process, we define the net and gross term for the utility of actions which allow us to evaluate achievements with relative terms within the ongoing process of utility maximization. Each process that improves utility must agree with a few several structural properties of optimal. We can define these properties over process due to the definition of the net and gross actions. Finally, we represent these properties with Value-Driven Landmarks, These Value Landmarks are domain-independent (can be applied in each task if sequential decisions or actions), and lead to better performance, sometimes, as you can see in the attached image, without a search at all.
The red circle emphasizes the tasks that solved without search since no plan that meets optimal properties as applicable. In real-world scenarios that involve budget thus is very likely to happen at some point during the search.
Oversubscription action planning with value landmarks - Empirical evaluation of the improving approach

Oversubscription action planning with value landmarks – Empirical evaluation of the improving approach

Stay tuned, we will keep update here the progress of our research and if you find that problem interesting, let us know, there is a lot of work and we will be more than happy to collaborate.

We will soon post a call for collaboration with some of our suggestions.

 

Goal-Oriented Bill Could Do it Better…

Hi,

Here is a new animation explainer I am working on. It comes to clear another point from our research on Oversubscription planning and the merits of value landmarks in a complex (day life) problems with sequential actions.

 

There are some tuning left; I would love to have your comments.
Thanks!

Oversubscription Planning (OSP)

Introduction to our research on oversubscription planning – OSP

Introduction

Oversubscription planning termed by D. Smith (2004), also referred to as OSP, deals with achieving a state with a high utility value, given a budget on total action cost. This formulation allows us to handle situations with under-constrained resources, which do not allow us to achieve all possible goal facts. OSP objective differs from classical planning objective, in which the objective is to find a cheap plan which achieves all goal facts. In optimal OSP and optimal classical planning, the tasks are further constrained to finding a path which achieves a state with maximal utility, and, to finding the cheapest cost path which achieves the goal, respectively.

Over the years, the theory and practice of classical planning have been studied and advanced much more intensively compared to OSP. Recent work (Mirkis & Domshlak 2013,2014; Domshlak & Mirkis 2015) made several contributions aiming at improving the scalability of OSP solvers. In particular, they developed a planner which exploits standard landmark discovery tools of classical planning, as well as abstractions for solving OSP problems. They showed how standard goal-reachability landmarks of certain classical planning tasks could be generated to represent achievements of valuable facts of the original OSP task. These landmarks compiled into the original OSP task to obtain an equivalent OSP task with a lower cost allowance, and thus with a smaller effective search space. In this research, we investigate approximation methods for OSP, aiming at extending the scope of the landmark-based approximations for OSP, as well as on improving the scalability of the state-space search driven by these approximations. We propose techniques which allow us to discover more informative landmarks than previous methods. Furthermore, our techniques are applicable for OSP tasks generalized to additive utility setting.

Starting with the most basic question of what additive utility function in OSP problem is, we point out the challenges in multi-valued planning tasks with additive utility setting. We discuss the relationships between state variables and different value assignments to a variable in successive states along a plan. We closely consider negative interactions between state variables with multi-valued (non-zero binary) utility setting, and we show how these negative interactions could occur in tasks with non-negative utility setting. We introduce a novel heuristic search approach to address additive utility OSP tasks which differs from the traditional automated action planning approach in the sense of interpretation of the objective of planning for valuable facts. With our focus on improving state-variables rather than on collecting valuable facts, we define the utility of actions. Investigation of planning tasks and their utility structure with the utility of action definition in hand allowed us to discover properties of optimal plans on the level of actions and sequences of actions. A synergistic combination of these properties allows handling OSP additive utility functions over multi-valued variables.

Continue reading →

Check out our paper in ICAPS 2018

Our paper on Value Driven Landmarks for Oversubscription Planning [1] is presented in the International Conference on Automated Planning and Scheduling  – ICAPS 2018

Oversubscription planning is the problem of choosing an action sequence which reaches a state with a high utility, given a budget for total action cost. Most previous work on oversubscription planning was restricted to only non-negative utility functions and 0-binary utility functions. While this restriction allows using techniques similar to partial satisfaction planning, it limits the expressivity of the formalism. In this paper, we address oversubscription planning with general additive utility functions over a finite-domain representation. We introduce the notions of net utility of an action, and of a gross positive action. Using these notions, we prove several properties about the structure of an optimal plan, which are then compiled into a classical planning problem. The landmarks of this classical planning problem are value driven landmarks of the original oversubscription problem, that is, they must occur in any action sequence which improves utility. An empirical evaluation demonstrates that these landmarks are more informative than previous state-of-the-art methods for landmark discovery for oversubscription planning, and lead to better planning performance.

[1] D. Muller and E. Karpas, “Value driven landmarks for oversubscription planning.,” in Icaps, 2018.
[Bibtex]
@inproceedings{muller:Karpas:icaps18,
title={Value Driven Landmarks for Oversubscription Planning.},
author={Muller, Daniel and Karpas, Erez},
booktitle={ICAPS},
year={2018}
}