Sesi 13

Agent Based Modeling

[TEORI]
 LIPUTAN
• DEFINISI MODEL BEBASIS AGEN
• PENGERTIAN AGEN DAN SIFAT-SIFATNYA
• PENGERTIAN EMERGENT
• ESAP: ENVIRONMENT, SENSOR, ACTUATOR,
PERFORMANCE,
• VARIASI AGEN
• AREA PENGGUNAAN MODEL BERBASIS AGEN
• PERBANDINGAN DENGAN MODEL LAIN
 Agent Based Modeling

• Computer simulation that represents individual actors in

a dynamic social system.

• ‘Agents’ represent heterogeneous individuals who

interact with each other and/or their environment based
on set rules.

• From these interactions, macro-scale behaviors may

emerge.
 Agent-Based Simulation
• Some of the simulated entities are agents
• Explicitly represents specific behaviors of specific individuals –
contrast with traditional macro-level aggregated representations
• Facilitates simulation of group behavior in highly dynamic
situations. Allows study of "emergent behavior"
• Parallel and distributed simulation for e.g., the crowd simulation
• Well-suited to populations of heterogeneous individuals
– vehicles (and pedestrians) in traffic situations
– actors in financial markets or social network
– consumer behavior
– humans and machines in battle fields
– people in crowds
– animals and/or plants in eco-systems
– artificial creatures in computer games
– enzymes, DNA, and mRNA in a bio cell
 Agents
• Agents are objects with attitude:
– Autonomous
• Operates without direct intervention of human or others
• Capable to modify their behaviour

– Proactive
• Actions depending on motivations generated from their internal state

– Social ability
• Interact with other agents with some kind of agent communication
language

– Reactive
• Agents perceive their environment, either physical or computer
• Respond in a timely fashion to changes that occur in it
 Agents
Agents have:
 Internal data representations
(memory or state)
 Means for modifying their internal
data representations (perceptions)
 Means for modifying their
environment (behaviors)

Mobil, bis, motor, kereta BUKAN agent karena tidak punya autonomi.
Kalau PLUS pengemudinya barulah agent karena bisa menjadi sarana
perpanjangan interaksi dari pengemudinya. Bagian-bagian dari
kendaraan menjadi activatornya dan mungkin juga sensornya.
 AGENT : formal definition
• Agent suatu entitas yang menerima input dari
lingkungannya (persepsi) dan bertindak (aksi) sesuai
dengan pengetahuan yang dimilikinya
• Atau secara abstrak, agent merupakan suatu fungsi dari
persepsi menjadi aksi
 Anatomy of an agent
An agent should consist of :

Actions
Sensors

The
ENVIRONMENT
Effector
Percepts
Percepts
 Spectrum of Agent Properties

intention goal
management management domain
belief capability knowledge
management capability
capability domains
…
language
agent
skills
model
decision
communication
making
capabilities
abilities
manipulation perception
mobility navigation abilities
skills
skills skills
 Definition
• Emergent behaviour is discernable macrobehaviour
resulting from interactions between agents
• Dynamic systems form patterns in time as well as space
• Emergent behaviours often display adaptation - they grow
smarter over time in response to the specific and changing
requirements of their environment
• The big question – if the behaviour of complex systems
can’t be predicted in detail, at what level and under what
circumstances can it be predicted?

13 Agent Based Modeling

PENGERTIAN EMERGENT
Kendaraan dan pedestrian dalam situasi trafik

Semua kendaraan dan orang berusaha

untuk bergerak maju ke arah tujuan
masing-masing. Dalam proses ini akan
terjadi posisi dan interaksi yang terus
selalu berubah setiap saat.
Setiap pengguna jalan harus waspada
tidak menabrak dan tertabrak;
Harus mencapai tujuan on-time dengan
selamat
PENGERTIAN EMERGENT

Setiap saat terjadi interaksi yang berubah-ubah

status dan posisinya. Banyak penjual dan banyak
pembeli yang saling ber-interaksi.
PENGERTIAN EMERGENT

Setiap saat terjadi multi-aktivitas interaktif antara

para pelaku dogfight yang pada saat berikutnya
merupakan akibat dari keputusan-tindakan yang
diambil.
Setiap pelaku mempunyai tujuan sama :
membunuh musuh, dan diri sendiri selamat
PENGERTIAN EMERGENT
INTERAKSI AGENT DENGAN
LINGKUNGANNYA
 Agents dan Lingkungan

• Agents adalah segala sesuatu yang dapat melihat/

mengartikan/ mengetahui (perceiving) linkunganya
melalui alat sensor (sensors) dan bertindak (acting)
melalui alat aktuator (actuators)
• Manusia sebagai agent : mata, telinga dan organ lainnya
sebagai sensors; tangan, kaki, mulut dan bagian tubuh
lainnya sebagai actuators
• Robot sebagai agent : kamera and pejejak infra merah
sebagai sensors; berbagai motor pengerak sebagai
actuators
• Software sebagai agent : tekanan pada keyboard, isi file
dan paket-paket pada jaringan sebagai masukan
sensors; tampilan pada layar, penulisan file dan
pengiriman paket jaringan sebagai keluaran actuators
 Agents dan Lingkungan
(Environments)

• Fungsi agent (f) adalah pemetaan dari urutan persepsi

(percept) menjadi tindakan (actions)
[f: P*  A]
• Program agent berjalan pada arsitektur fisik untuk
menghasilkan fungsi agent (f)
• agent = architecture + program
 Vacuum-cleaner world

• Environment: square A and B

• Percepts: [location and content] e.g. [A, Dirty]
• Actions: left, right, suck, and no-op

function REFLEX-VACUUM-AGENT ([location, status])

return an action:
if status == Dirty then return Suck
else if location == A then return Right
else if location == B then return Left
Vacuum-cleaner world - lanjut

Percept sequence Action

[A,Clean] Right
[A, Dirty] Suck
[B, Clean] Left
[B, Dirty] Suck
[A, Clean],[A, Clean] Right
[A, Clean],[A, Dirty] Suck
… …
 Konsep Rasionalitas
• Rasional tergantung pada 4 hal :
– Kemampuan yang terukur,
– Pengetahuan lingkungan
sebelumnya/terdahulu,
– Tindakan  yang mengarah kepada tujuan
– Urutan persepsi (sensors).

• DEF: Untuk setiap urutan persepsi yang mungkin, rational agent

harus memilih tindakan yang diharapkan dapat memaksimalkan
kemampuan dengan memberikan bukti yang dihasilkan dari urutan
persepsi dan pengetahuan yang dimiliki oleh agent.
 Konsep Rasionalitas
• Rationalitas  kemahatahuan (omniscience)
– Agent perlu mengetahui sebagian akibat yang terjadi
dari suatu tindakan, tapi tidak mengetahui semua
kemungkinan.
• Agent dapat bertindak sesuai dengan yang
diharapkan untuk memodifikasi persepsi.
• Agen perlu mendapatkan informasi yang
berguna (pengumpulan informasi dan
eksplorasi)
• Agent dikatakan autonomous, jika perilakunya
ditentukan oleh pengalamannya sendiri (dengan
kemampuan untuk belajar dan beradaptasi)
 Konsep Rasionalitas
• Rational agent adalah agent yang melakukan sesuatu yang
benar
– Setiap kolom pada tabel (Vacuum-cleaner world)
diselesaikan/dikerjakan dengan benar
• Apakah sesuatu yang benar?
– Agent yang paling sukses/ berhasil
– Mengukur kesuksesan/ keberhasilan?
• Pengukur kemampuan haruslah objektif (contoh : Vacuum-
cleaner world)
– Jumlah debu yang dapat dibersihkan pada suatu kurun
waktu tertentu
– Besarnya konsumsi listrik yang diperlukan
– Besarnya noise yang dihasilkan
–…
 ESAP
• To design a rational agent we must specify its
task environment.
• ESAP description of the environment:
– Environment
– Sensors
– Actuators
– Performance
• Beware that all performance issues have
units and must be quantified
 ESAP – contoh 1
Situasi : Pasar tradisional
• Environment: lapak/kios, penjual/pembeli,
barang jualan, carry-boy, pintu/tangga, dll
• Sensors: mata, telinga, perabaan tangan
• Actuators: mulut/suara, tangan/kaki
• Performance: pembeli: semua barang didapat
dengan harga murah. Penjual: barang terjual
sebanyak mungkin dengan untung setinggi
mungkin dan pembeli puas.
 ESAP – contoh 2
Situasi : Peperangan di udara
– Environment: angkasa, pesawat tempur, awan,
dll
– Sensors: mata penerbang, layar/dashboard
pesawat,
– Actuators: Meriam, sayap/ekor pesawat,
tombol-tombol pesawat, tangan/kaki pilot
– Performance measure: Jumlah pesawat musuh
yang dijatuhkan, waktu cepat, dan biaya sedikit
 Agent types; goal-based

• Tujuan-tujuan tertentu dapat di capai dengan cara-cara berbeda.

– Beberapa lebih baik, memiliki manfaat yang lebih tinggi.
• Fungsi utililas memetakan urutan kedudukan ( a sequence of
states) dengan angka real.
• Meningkatkan tujuan-tujuan :
– Memilih tujuan dari tujuan-tujuan yang berbenturan
– Memilih dengan tepat beberapa tujuan memiliki kemungkinan berhasil.
 Agent types; utility-based

• Agent membutuhkan tujuan untuk mengetahui situasi mana yang

diharapkan.
– Akan menjadi sulit ketika urutan yang panjang dari tindakan-tindakan
(actions) dibutuhkan untuk mencari tujuan.
• Typically investigated in search and planning research.
• Major difference: future is taken into account
• Is more flexible since knowledge is represented explicitly and can
be manipulated.
 Agent types; learning

• Semua program-program agent terdahulu mendeskriopsikan

metoda untuk memilih tindakan-tindakan (actions).
– Yet it does not explain the origin of these programs.
– Learning mechanisms can be used to perform this task.
– Teach them instead of instructing them.
– Advantage is the robustness of the program toward initially unknown
environments.
Agent types; learning - lanjut

• Learning element: introduce improvements in performance element.

– Critic provides feedback on agents performance based on fixed
performance standard.
• Performance element: selecting actions based on percepts.
– Corresponds to the previous agent programs
• Problem generator: suggests actions that will lead to new and
informative experiences.
– Exploration vs. exploitation
 Area Penggunaan ABM

Pada dasarnya setiap situasi sosial, bisnis

atau engineering yang melibatkan dua atau
lebih pelaku yang saling berinteraksi dengan
tujuan masing-masing berbeda atau
berkoordinasi – kooperasi - kompetisi –
negosiasi – [ataupun planning] dapat
menggunakan pemodelan berbasis agen
Beberapa contoh pemodelan berbasis agen
(Axelrod-Tesfatsion)

ABM is well suited for social science objective, with two properties:
(1)the system is composed of interacting agents; and
(2)the system exhibits emergent properties, that is, properties arising from the
interactions of the agents that cannot be deduced simply by aggregating the
properties of the agents.
•When the interaction of the agents is contingent on past experience, and especially
when the agents continually adapt to that experience, mathematical analysis is
typically very limited in its ability to derive the dynamic consequences. In this case,
ABM might be the only practical method of analysis.
•ABM begins with assumptions about agents and their interactions and then uses
computer simulation to generate “histories” that can reveal the dynamic
consequences of these assumptions.
•ABM can investigate how large-scale effects arise from the micro-processes of
interactions among many agents.
 COMPARISON OF ABS and
SYSTEM DYNAMICS
ABS System Dynamics
Process: Inductive Process: Deductive

Unit of analysis: agent / Unit of analysis: feedback loop

individual / structures
Bottom-up approach DES relies upon probability
distributions “Top down” (DES
Focus: Exploratory research Focus: Confirmatory research
 Agents versus Objects

• Like objects, agents encapsulate state and

behaviour
• But objects are essentially passive
– They have no choice over interaction
– They are invoked by other objects
• Agents decide whether to perform the
desired operation
• Agents are autonomous
 Agents versus Objects:
Modeling
• Objects provide a valuable way to model the world
• But agents provide a more natural representation of
real-world systems in which different individuals
interact according to their own agendas and
priorities.
• They can come together to achieve overarching
objectives that might not, or not as easily, be
achieved by the individuals alone
 Agents versus Objects
• When agent goals are closely aligned, and if agents
are benevolent and honest, then the resulting
systems may resemble an object-oriented system.
• The object-oriented paradigm may be adequate, but
is unlikely to be flexible.
• Agents may be constructed using object
technology.
• Agents typically run in their own thread of control,
while standard object systems have one thread.
 MEASURED RESPONSE:
MATHEMATICAL MODELS AS
DETERMINANTS OF AGENT
BEHAVIORS

• Agent based Computational Environment

– Genomic Computing
• Behavior and Mobility Modeling
• Epidemiological Modeling and Calibration
• Person in the Loop
 MEASURED RESPONSE:
EPIDEMIOLOGICAL MODEL
AS CALIBRATOR OF ABS
S I R
• Susceptible-Infected-Recovered (SIR) model for population
N=S+I+R with no disease mortality.
• Mass action transmission process, rate linear recovery
rate 
 ABS AS CONTINUOUS
EXPERIMENTATION
• Simulation as a persistent process
• Continuous availability of a virtual, or synthetic,
environment for decision support (ex: artificial labor
market)
• Continuous, “near real time” sensor data from real world
counterpart (via data warehouse)
• “Parallel worlds” interaction
• Agents in the ALM developed using existing OR/MS models
as data mining tools from the data warehouse
• Calibrate the ALM using existing OR/MS models
• ABS as test bed for OR/MS models
 ABS AS CONTINUOUS EXPERIMENTATION:
PARALLEL WORLDS
Time Simulation Loop
Compression
Decision Support Loop
Near exact replica
of the “real” world

Synthetic Assess Real World

Environment Environment
Behavior
modeling,
SEAS architecture SCM
demographics,
Supports millions of
ERP Data collection,
Artificial agents association,
and calibration CRM trends, and
Data parameter
Warehouse estimation
Learn: Explore, Experiment,
Implement
Analyze, Test, Predict DECISION

XML Interfaces
UNIX/ORACLE The user(s) can
seamlessly switch
Real World and between real and
Simulation Databases virtual worlds through
an intuitive user
interface.
 ABS AS CONTINUOUS
EXPERIMENTATION

DATA WAREHOUSE

CALIBRATING AGENTS:
OPTIMIZATION MODEL:
OR/MS models to
“Where are the best
Validate Market
PROGRAMMING AGENTS: locations for Recruit
Behavior
Data Mining using Stations?”
Econometric Models,
Neural Networks, etc
to Specify Preferences
DEMAND MODEL:
“What will be the recruit
pool by race, gender,
and location next year?”
ARTIFICIAL LABOR MARKET
 ABS AS CONTINUOUS
EXPERIMENTATION:
ARTIFICIAL LABOR MARKET

• Agent-based simulation designed to

capture the dynamics of a labor market
• Agents represent individuals, or cohorts,
in the labor market
• Humans play role(s) of organizations
• Agents programmed with “rules of
engagement” + genetic structure